Int. J.

Production Economics 165 (2015) 247–259

Contents lists available at ScienceDirect

Int. J. Production Economics

journal homepage: www.elsevier.com/locate/ijpe

Insights from hashtag #supplychain and Twitter Analytics: Considering

Twitter and Twitter data for supply chain practice and research
Bongsug (Kevin) Chae
Department of Management, College of Business Administration, Kansas State University, United States

art ic l e i nf o a b s t r a c t

Article history: Recently, businesses and research communities have paid a lot of attention to social media and big data.
Received 1 March 2014 However, the field of supply chain management (SCM) has been relatively slow in studying social media
Accepted 28 December 2014 and big data for research and practice. In these contexts, this research contributes to the SCM community
Available online 5 January 2015
by proposing a novel, analytical framework (Twitter Analytics) for analyzing supply chain tweets,
Keywords: highlighting the current use of Twitter in supply chain contexts, and further developing insights into the
Supply chain management potential role of Twitter for supply chain practice and research. The proposed framework combines three
Twitter methodologies – descriptive analytics (DA), content analytics (CA) integrating text mining and sentiment
Data analytics analysis, and network analytics (NA) relying on network visualization and metrics – for extracting
Network analytics
intelligence from 22,399 #supplychain tweets. Some of the findings are: supply chain tweets are used by
Content analytics
different groups of supply chain professionals and organizations (e.g., news services, IT companies,
Big data
Social media analytics logistic providers, manufacturers) for information sharing, hiring professionals, and communicating with
Application Programming Interface (API) stakeholders, among others; diverse topics are being discussed, ranging from logistics and corporate
social responsibility, to risk, manufacturing, SCM IT and even human rights; some tweets carry strong
sentiments about companies' delivery services, sales performance, and environmental standards, and
risk and disruption in supply chains. Based on these findings, this research presents insights into the use
and potential role of Twitter for supply chain practices (e.g., professional networking, stakeholder
engagement, demand shaping, new product/service development, supply chain risk management) and
the implications for research. Finally, the limitations of the current study and suggestions for future
research are presented.
& 2015 Elsevier B.V. All rights reserved.

1. Introduction media as a medium for marketing, customer engagement, recruit-

ment, sales forecasting, and educating.
Recently, businesses and research communities have paid a lot of The field of supply chain management (SCM) has been relatively
attention to social media (Aral et al., 2013; Harris, 2014; Kalampokis slow in identifying the potential role and use of social media for
et al., 2013). Businesses are primarily interested in the role and use of research and practice. While there are a growing number of studies
social media for marketing purposes, including brand management and reports on the use of data and analytical capabilities for SCM
and product and service promotion. Research communities, made up (Chae and Olson, 2013; Hazen et al., 2014; Trkman et al., 2010), the
of members from diverse academic fields, have been investigating focus has generally remained on traditional data sources (e.g.,
the potential use of data (also referred to as “big data”), created and enterprise resource planning) and analytical techniques (e.g., opti-
stored through social media technologies or platforms, to develop mization algorithms), and their use for and impact on supply chain
new insights in different areas, including stock price predictions, planning and execution. Nevertheless, a small number of seminal
prevention of epidemics, early event monitoring, election predictions, articles on the use and potential role of social media in the supply
crisis management and humanitarian relief, brand management, chain have been published in the academic literature and industry
public relations, information diffusion, and public opinions (Arias press (Casemore, 2012; O’Leary, 2011). Also, there are recent calls for
et al., 2014; Hughes and Palen, 2009; Inauen and Schoeneborn, 2014; the utilization of big data in the SCM field (e.g., Huang et al., 2014;
Williams et al., 2013). Industry communities (e.g., companies, profes- Smart et al., 2014; Waller and Fawcett, 2013). Big data created in
sionals, associations) have explored many potential uses of social manufacturing alone is close to 2 exabytes (2
10006) in 2010, which
is more than in any other sectors (McKinsey&Company, 2011). While
there are many sources (e.g., sensors, transaction logs) for big data,
E-mail address: kevinbschae@gmail.com there is an explosive growth of Web 2.0 applications (e.g., Twitter,

http://dx.doi.org/10.1016/j.ijpe.2014.12.037
0925-5273/& 2015 Elsevier B.V. All rights reserved.
248 B.(K.) Chae / Int. J. Production Economics 165 (2015) 247–259

Facebook, blogs) in recent years, and, thus, social media largely (3) What are the characteristics of those Twitter users who are
contributes to the emergence of these big data. discussing supply chain-related topics?
A recent survey of industry leaders and companies (Cecere, (4) What are the sentiments of supply chain tweets? What types
2012) reports that one in three companies is evaluating a big data of supply chain tweets tend to contain sentiment?
initiative for SCM. The companies in the survey express confidence
in their ability to manage traditionally structured and transactional These findings help us acquire greater insights into the use and
data, including product traceability data, geo-location and mapping potential role of Twitter for supply chain practices (e.g., stake-
data, and supply chain visibility. However, the survey notes that the holder engagement, hiring supply chain professionals, new pro-
companies are least capable of using social media data for supply duct/service development, supply chain risk management), and to
chain intelligence. Another industry survey (Natoli, 2013) shares a further discuss research implications.
similar story: while logistics providers, manufacturers, and retailers We organized the paper as follows: Section 2 presents a brief
are leveraging traditional supply chain data (e.g., POS), only one review of Twitter and the use of Twitter in other areas; Section 3
percent of those surveyed reported any use of social media data for proposes a framework of Twitter Analytics (TA), along with some
supply chain planning. An industry expert echoes these situations, relevant studies; Section 4 applies the TA framework to the
“We know social media will transform supply chain processes, analysis of 22,399 supply chain-related tweets and metadata; in
but we just don't know how exactly and where to start and why.” Section 5, we present a broad range of intelligence from descrip-
(Mann, 2014). tive, content, and network analytics techniques, and further
In these contexts, we pay attention to one particular social discuss them in the context of supply chain; in Section 6, based
media platform, Twitter. We chose Twitter because, among diffe- on the intelligence, we develop greater insights into the use and
rent social media technologies, Twitter has become the fastest potential role of Twitter for supply chain practice and discuss
growing social platform, ahead of Facebook and Google þ research implications. The final section presents the limitations of
(Bennett, 2013). Currently, over 270 active Twitter users generate the current study and makes suggestions for future research.
500 million tweets per day.1 Customers follow products, services
and brands, and discuss them in Twitter (Webster, 2010). Another
important reason is that, unlike Facebook data, Twitter data could
be considered “open”. Thus, research and business communities 2. Background: Twitter and its impacts
can access Twitter data using Twitter Application Programming
Interface (API) (Twitter, 2013), which has offered them opportu- Twitter, along with Facebook and Youtube, represents the rapid
nities to access the data in an unprecedented scale and size, growth of Web 2.0, or social media applications, in recent years. Since
and to analyze such data for challenging problems in diverse its inception in 2006, this microblogging application has been the
domains. fastest growing social media platform. Over 75% of the Fortune
The objective of this paper is to improve our understanding of Global 100 own one or more Twitter accounts at the corporate level
Twitter, in particular (and social media in general), in supply chain and for their specific brands (Malhotra et al., 2012). People and
contexts. More specifically, the paper proposes a novel, analytical organizations form dynamic communities through following and
framework (Twitter Analytics) for analyzing supply chain tweets, being followed by others, and information is disseminated rapidly, as
highlighting the current use of Twitter in supply chain contexts, recently exemplified by the stock market's response to a fake tweet
and further developing insights into the potential role of Twitter from the account of the Associated Press.
for supply chain practice and research. As noted above, there are A tweet is a message of up to 140 characters. Three types of
few research articles and published industry cases on the topic of tweets are found in Twitter: original tweets, replies, and retweets.2
social media and supply chain. While it is imperative to under- Original tweets appear in the sender's profile page and Home
stand social media and social media data in supply chain contexts, timeline. Such an original tweet can be retweeted by other users.
as echoed by industry surveys and academics (e.g., Cecere, 2012; Also, Twitter users can join the conversation by @replying to others
Mann, 2014; Smart et al., 2014), it is clear that there is a lack of and by retweeting. These tweets are traceable. Twitter offers API to
understanding of this important topic. To our knowledge, a practi- researchers, practitioners, and organizations that are interested in
cal methodology or framework for analyzing social media data, collecting and analyzing tweets (Twitter, 2013). Search and Stream-
related to supply chain practice (and research), is not readily ing APIs allow researchers and businesses to collect Twitter data
available. Thus, this research is expected to make important using different types of queries, including keywords and user
contributions to the SCM community. profiles. While Facebook makes only the data in group and fan pages
Twitter Analytics (TA) combines three types of analytics (or (e.g., Global Supply Chain Group in Facebook) open to the public,
research techniques) – descriptive analytics (DA), content analytics Twitter data is considered “open data”.
(CA), and network analytics (NA) – each of which focuses on In recent years, Twitter data has become one of the most
different dimensions in the analysis of Twitter data. This framework popular information sources for practical applications and academic
is applied to 22,399 tweets and metadata, which were collected research. There are numerous examples of practical applications of
using hashtag #supplychain, the most prevalent supply chain- Twitter data, ranging from stock forecasting (Arias et al., 2014;
related Twitter hashtag. The result is a large amount of intelligence Feldman, 2013), through real-time event and trend analysis using
regarding supply chain professionals and organizations and their machine learning algorithms (Dickey, 2014), brand management
Twitter use. The findings are discussed in an organized format (Malhotra et al., 2012), to crisis management (Wyatt, 2013). It is
around four general questions: expected that there will be a rapid growth of Twitter data use for
many other applications, including public safety, market prediction,
(1) What are the characteristics of supply chain tweets? Are there and humanitarian assistance and relief (Dataminr, 2014).
any patterns of communication and information diffusion? Also, Twitter has already impacted academic research in many
(2) What supply chain topics or contents are shared in Twitter? fields, including finance (Bollen et al., 2011), healthcare (Park et al.,
Are there any prevalent topics or contents? 2013; Terry, 2009), journalism (Lasorsa et al., 2012), information

2
https://support.twitter.com/articles/
1
https://about.twitter.com/company. 119138-types-of-tweets-and-where-they-appear.
 B.(K.) Chae / Int. J. Production Economics 165 (2015) 247–259 249

systems (Aral et al., 2013), politics (Gayo-Avello, 2012), marketing Stieglitz, 2013). This user information is closely related to the
(Jansen and Zhang, 2009), communication (Zappavigna, 2011), and outcome of centrality (or popularity) analysis from network analytics
psychology (Dodds et al., 2011). In the SCM area, the study of (NA). Finally, a large portion of tweets contain one or more URLs in
Twitter (social media and big data in general) is rare. As an their texts. URLs could be news releases, reports, articles, and more.
exception, O’Leary (2011) suggested the use of Twitter for detect- Thus, analyzing URLs can reveal what topical interests and informa-
ing supply chain events and facilitating frequent communication tion are considered important among Twitter users.
among supply chain partners. Twitter's consequences are expected While only these three types of analyses are introduced, other
to be far-reaching in research and to inspire other academic fields. descriptive analyses and metrics are certainly possible and should
be used for different problems. However, the use of too many
metrics is likely to cause information overload and confusion,
3. A framework of extracting intelligence from Twitter data rather than intelligence. When using DA, practitioners and
researchers should carefully consider a selective list of analyses
Collecting Twitter data (tweets and metadata) begins with and metrics according to the questions they are trying to address.
identifying the topic of interest using a keyword(s) or hashtag(s), Also, the change of analyses and metrics is expected when DA is
and requires the use of APIs (e.g., Streaming). This API method used for other social media data (e.g., Facebook).
allows acquiring 1% of publicly available Twitter data. Twitter data
is also available through data providers (e.g., GNIP, DataSift), also
known as Twitter Firehoses, which can deliver 100% of Twitter 3.2. Content analytics (CA)
data based on criteria. This is an ideal, but very costly, option.
Other social media platforms also provide their API services. For Social media data are primarily texts and thus “unstructured” in
example, Facebook offers Graph API. nature. Thus, it is necessary to use content analytics (CA), which
While Twitter (and social media) data collection can rely on refers to a broad set of natural language processing (NLP) and text
standardized API services, the analysis of such collected data mining methods, for extracting intelligence from Web 2.0 (Chau
becomes challenging because the data are less structured (e.g., and Xu, 2012). A tweet's text is informal and composed of a short
texts, informal expressions) and more enriched (e.g., user profiles, list of words, hashtags, URLs, and other information. Thus, a careful
follower, hashtags, URL) (Daniel et al., 2010) than traditional data consideration of text cleaning and processing is a prerequisite for
(e.g., sales data) found in corporate databases, and the analytical intelligence gathering. Also, tweets (e.g., “#SupplyChain of cheap
framework or methodology is not readily available. The use of clothing stained with the blood of #Bangladeshi workers – http://
diverse research methods and metrics is necessary to extract nydn.us/181XT2M – via @nydailynews”) contain not just informa-
intelligence from the highly enriched and unstructured social tion, but also opinions. Thus, advanced text mining techniques, such
media data (Chau and Xu, 2012; Daniel et al., 2010; Fan and as sentiment analysis, are the key for extracting such opinions.
Gordon, 2014). With this in mind, the rest of this section focuses Text mining and machine learning algorithms are important
on developing an analytical framework encompassing such components of CA. Text mining transforms unstructured texts (or
research methods and metrics for extracting intelligence from documents) into formatted data (or documents), using such
Twitter data. It consists of three methodologies from different techniques as tokenization, n-grams, stemming, and removing
intellectual backgrounds: descriptive analytics (DA), content ana- stop words (unnecessary words) (c.f., Weiss et al., 2005). Those
lytics (CA), and network analytics (NA) with some relevant transformed texts can be used for text summarization, key word
metrics3 (see Fig. 1). analysis, word frequency analysis, and text clustering, using
machine learning algorithms, such as clustering and association
analysis. While CA is found in supply chain research and practice
3.1. Descriptive analytics (DA) (Georgi et al., 2010; Seuring and Gold, 2012; Vallet-Bellmunt et al.,
2011), the approach has been manual or semi-manual, primarily
Twitter data contain a large amount of information, including through human interpretations. CA in TA relies on automatic text
tweets and metadata (e.g., user information). DA focuses on descrip- processing techniques and algorithms, due to the big data nature
tive statistics, such as the number of tweets, distribution of different of Twitter data.
types of tweets, and the number of hashtags. Descriptive statistics are Word analysis is a starting point in CA. It includes term
widely used in SCM research and practice. For example, researchers frequency analysis, document summarization, and clustering. Term
always report descriptive statistics about the survey they conducted. frequency (TF) is widely used in information retrieval. TF can be
The difference lies in the number of metrics. While a small number of combined with n-gram, which helps to identify key phrases from
metrics (e.g., sample size, response rate, responder profile) are used the documents. Overall, these analyses help identify topics of
for the survey data, the enriched nature of Twitter data enables discussion, which also indicate different usages of Twitter according
intelligence extraction, using a large set of metrics regarding tweets, to supply chain contexts. Those identified topics are useful for
users, hashtags, URLs, etc. document-level analysis, using unsupervised machine learning
A simple, but broad, view of tweet data is a pre-requisite for algorithms (e.g., clustering). Document-level clustering can categor-
detailed analyses. Tweet metrics aim to present a simplistic, but ize documents, which helps enable detailed analysis of those
essential, picture of the data, using several metrics (number of documents in terms of their categorization. Second, hashtags are
tweets, word counts, @user per tweet, number of hashtags, etc.) an important component of tweets, as they are equivalent to areas
(Bruns and Stieglitz, 2013). These are the gateway for many other or fields of interest used to categorize academic papers (or
metrics. Second, knowing who tweets, replies, and retweets is researchers) or tags used to categorize blog postings. In the above
important both for practitioners looking for business value from example of a tweet, #supplychain and #Bangladeshi are hashtags.
Twitter and researchers studying a phenomenon. The user metrics Hashtag analysis includes frequency analysis and association rule
show most active/visible users, groups of users in terms of their mining. Frequency analysis shows the popularity of hashtags.
activities, and other useful user-related information (Bruns and Association rule mining discovers relationships between hashtags.
These relationships explain how different areas of business practice
3
The list of those metrics in Fig. 1 is not exhaustive. It includes only some of (logistics, corporate social responsibility, risk management, etc.) are
the potential metrics. intersecting in Twitter.
250 B.(K.) Chae / Int. J. Production Economics 165 (2015) 247–259

Fig. 1. A proposed framework of extracting intelligence from Twitter (Twitter Analytics).

While word and hashtag analyses focus on finding facts in the centrality focuses on those nodes adjacent to a focal node,
tweet content, finally, sentiment analysis (or opinion mining) is “betweenness centrality” includes distant paths of the focal node.
primarily interested in extracting subjective information (e.g., emo- While centrality analysis mainly focuses on individual nodes (or
tion, opinions) in tweets (Feldman, 2013; Pang and Lee, 2008). users in Twitter), community analysis explores network-level
Sentiment analysis can be performed at two levels: (1) the entire characteristics. For example, network density represents the por-
tweets; and (2) several clusters of tweets in terms of themes. The tion of all possible connections between nodes, and, thus, it is a
task is to classify each tweet as negative, neutral, or positive. The measure of network cohesion (Wasserman and Faust, 2005).
above tweet about supply chain and Bangladeshi workers is an Modularity is a measure of how strongly the network is divided
example of tweets with negative sentiment. Thus, analysis of the into modules. Modularity analysis identifies specific communities
entire tweet data can reveal the overall sentiment of a keyword or a from the network through visualization.
hashtag. Also, it is possible to identify different clusters of tweets in In addition to these analyses and metrics, there is a myriad of
terms of themes. Tweets with #supplychain can be further clustered network concepts, analyses, and metrics. Thus, a well-framed
into sub-groups. Sentiment analysis can be conducted for each research question can be helpful for selecting a manageable list of
cluster of tweets. There are different techniques and tools for analyses and metrics when using descriptive analytics (DA). Proper
sentiment analysis (Pang and Lee, 2008). Thus, for high analysis changes in analysis techniques and metrics should be made for other
accuracy, the choice of sentiment analysis technique/tool should be social media data. For example, Facebook data enables the layout of
made carefully, while considering the key characteristic of a tweet, different networks (e.g., personal, group, fan page, timeline).
which is short and informal.

3.3. Network analytics (NA) 4. Research method

Twitter users engage through @reply and retweet. As a result, it The ideal approach to demonstrate the proposed framework
is possible to extract network information from Twitter data using and to gain insights into the use and potential role of Twitter in
the techniques and metrics in network theory, which is increasingly supply chain contexts, is to collect the entire Twitter data for a
used in many academic disciplines (Burt et al., 2013), including SCM certain period and extract relevant intelligence. However, given
(Borgatti and Lin, 2009; Carter et al., 2007; Galaskiewicz, 2011; Kim the vast amount of Twitter data (about 500 million tweets per
et al., 2011). Nodes (e.g., Twitter users) and edges (e.g., relation- day), all studies using Twitter data need a data sampling process,
ships) are two basic terms in the theory. Network topology refers to which relies on hashtags and/or keywords to identify relevant
a layout of the nodes and the edges based on the information of information. To collect tweets discussing supply chain-related
reply and retweet in Twitter. This network visualization uncovers topics and events from Twitter, we initially conducted a series of
patterns in interactions among users. Various network metrics (e.g., key word and hashtag searches, including “supply chain”, “SCM”,
average path length) provide the detailed description of such a “logistics”, “supply chain management”, and so on. This led us to
network. Using Twitter data, there could be two kinds of topological find out that #supplychain is the most prevalent hashtag used by
network: friendship network and @reply (or mention) network. supply chain professionals and organizations. Our formal data
Friendship networks can be constructed based on the information collection was conducted between February 5 and April 10, 2013.
of follower and following. Also, the conversation using @reply The final dataset includes 22,399 tweets with the hashtag #sup-
creates interpersonal relationships among Twitter users. plychain and their metadata.
In addition, network theory offers centrality analysis, which
uses node-level metrics, such as degree and betweenness cen- 4.1. Descriptive analytics (DA)
trality, revealing influential actors in the network. Degree central-
ity, a key metric, explains who has the most ties (or degrees) to Descriptive analytics of Twitter (and social media) data is
others in the network (Wasserman and Faust, 2005). While degree considered one of the key areas of interest for businesses and
 B.(K.) Chae / Int. J. Production Economics 165 (2015) 247–259 251

Fig. 2. Most active users vs. most visible users.

Table 1
Detailed word analyses in clustered documents.

Logistics fre Sustainability fre Manufacturing fre Risk fre Software fre

logistics 2551 sustainable 118 manufacturing 835 risk 536 mobile 103
jobs 1070 csr 61 risk 206 execs 193 android 96
careers 1014 food 29 reshoring 196 manufacturing 182 cyber 90
freight 262 csrwire 32 sourcing 149 part 125 portal 84
software 251 improving 36 part 128 planning 123 tracking 78
trucking 242 greenbiz 32 strategic 127 strategic 122 socialmedia 70
transportation 165 green 29 planning 124 deloitte 102 routes 66
shippers 112 procurement 27 mfgexecutive 116 risks 110 optimize 55
logisticsviewpt 96 ecodesk 27 ajsweatt 115 mfgexecutive 107 app 44

researchers. There are companies (e.g., twitonomy) with commer- https://community.kinaxis.com, http://paper.li (SCM online maga-
cial application services, which offer some descriptive metrics. To zine), http://www.ebnonline.com (SCM online magazine), http://
perform DA, we used script languages (Bruns and Burgess, 2011) www.csrwire.com (news related to corporate social responsibility),
and multiple statistical and data mining techniques. Script lan- http://www.analytiqa.com (business intelligence company), http://
guages were needed to extract such information as users and www.industryweek.com, http://www.scmr.com (supply chain man-
hashtags from the dataset of tweets. Statistical and data mining agement review), and http://www.inboundlogistics.com (logistics
techniques helped develop and visualize descriptive statistics. magazine), among others.
Tweet statistics: Among 22,399 tweets, original tweets, retweets,
and @replies account for 58% (12,920), 28% (6369), and 14% (3110), 4.2. Content analytics (CA)
respectively. We found more than 3800 different hashtags in the
tweets, ranging from popular SCM hashtags (e.g. #logistics, #man- CA involved three types of analysis: word analysis, hashtag
ufacturing) to others, such as #slave and #telehealth. Over 16,000 analysis, and sentiment analysis. Word analysis further included
tweets (70% of the tweets) contained more than 2 hashtags, indicat- term frequency analysis and clustered document-level analysis.
ing that the majority of tweets are intersecting multiple areas of Hashtag analysis was based on hashtag frequency analysis and
interest. For example, tweets contain two or more hashtags: #sup- association analysis of different hashtags. Finally, sentiment ana-
plychain, #CSR (corporate social responsibility), and #ethics. An lysis was conducted at two levels: the subjectivity of (1) the entire
example is “Too often, companies do NOT know the source of tweets, and (2) those categorized or clustered tweets according to
materials in their supply chain #CSR #Ethics #supplychain”. five top themes (CSR, Risk, Logistics, Manufacturing, and IT). CA
User analysis: We found 4313 unique users in the dataset. This relied on natural language processing (NLP) techniques, associa-
means each user sends out, on average, 5.19 tweets: 3 original tweets, tion algorithms, document clustering techniques, and opinion
1.48 retweets, and 0.71 @replies per user. Active users are calculated mining. Detailed discussions of these topics can be found in the
based on the number of tweets (original tweetsþretweetsþ @replies). NLP and sentiment analysis literature (Bird et al., 2009; Feldman,
The visibility of users can be calculated by the number of @replies 2013; Manning et al., 2008; Pang and Lee, 2008).
received (¼@replies receivedþretweets received) (Bruns and Burgess, Word analysis: Most popular words in tweets were logistics
2011). Fig. 2 shows the most active users and the most visible users: (found in 3961 tweets), jobs (3874), manufacturing (1309), careers
the most active users are not necessarily the most visible users. We (1160), risk (744), procurement (719), warehouse (614), software
also visualized the activity of most visible users. The result suggests (605), sustainability (602), import (549), export (544), freight (526),
that highly visible users tend to be active users as well. Major logistics sourcing (525), global (515), operations (503), retail (470), cloud
providers (e.g., UPS), manufacturers (e.g., Unilever), and retailers have (314), data (276), re-shoring (276), visibility (238), and security
Twitter accounts. They are visible but not active in our data because (219), among others. We further clustered the tweets in terms of
most of their tweets may not contain #supplychain. five popular themes – logistics, sustainability, manufacturing, risk,
URL analysis: URLs are popular in tweets. 89% (19,890) of the total and software – in the dataset, and then conducted word analysis.
tweets contained one or more URLs. 11,362 different URLs were The detailed findings are shown in Table 1.
found from these tweets. Especially active and visible users include Hashtag analysis: 3839 different hashtags were found in the
URLs in their tweets: almost all tweets from those users contain at tweets and they appear 62,575 times. The most popular hashtags
least one URL. Top URLs include companies' webpages, SCM online include #supplychain, #jobs, #tweetjobs, #logistics, #careers,
newspaper sites, and articles about manufacturing leadership and big #manufacturing, #sustainability, #export, #import, #brian-
data. Top URL domains include http://www.tweetmyjobs.com (job shube, #procurement, #equity, #manufacturing, #csr, #mobile,
announcements), http://www.linkedin.com, http://news.google.com, #3PL, #freight, #sourcing, #Android, #IT, #trucking, and
252 B.(K.) Chae / Int. J. Production Economics 165 (2015) 247–259

#distribution, among others. Two hashtags were most common large and sparse network. Despite this, the average path length is
in the tweets. On average, there were 2.79 hashtags per tweet. 5.82, indicating that, on average, everyone is about six nodes away
This warrants association analysis among hashtags. The analysis from each other. This seems to be due to a number of popular
shows that the items (#logistics and #supplychain) are most nodes (or hubs) connecting different groups of nodes, which are
popular, appearing in 5.6% (1254 tweets) of the total tweets sparse (Galaskiewicz, 2011; Watts, 2003).
(22,399). Centrality analysis (node-level metrics): Among a few node-
Sentiment analysis: There are different tools and algorithms for level network metrics, in-degree is a simple measure of a node's
sentiment analysis. We chose SentiStrength because the sentiment connectedness with others and can be an indicator of a user's
analysis algorithm is designed for informal texts such as tweets popularity. Companies (e.g., @deloitte, @toyotaequipment) and
(Thelwall et al., 2011). Fig. 3 shows the sentiments at the entire industry presses (e.g., @ILmagazine) in the SCM field draw many
dataset level. Many tweets appear to be neutral (neither positive replies or mentions, as indicated by their high in-degree values.
nor negative), as indicated by a high portion (67%) of tweets with Some individual users (e.g., @Lcecere) are noticeable. In addition,
the score 0. Another large portion (28%) of tweets scores either  1 many users mention CSR-related industry presses (e.g., @Guar-
or þ1, which is relatively neutral. However, some tweets contain dianSustBiz, @CSRwire) and individuals (e.g., @AmanSinghCSR) in
either strongly negative or positive sentiment. Table 2 shows some their tweets (Table 4).
exemplar tweets with relatively strong sentiments. Many of those with high connections are shown to be in
Next, we separated the tweets into five clusters, using such communication paths among other users. For example, @kinaxis,
themes as CSR, Risk, Logistics, Manufacturing, and IT. The outcome a SCM IT solution company, is in connection paths between some
is introduced in Fig. 4. The pattern that emerged from the hubs, including @Ilmagazine, @Lcecere, etc. Looking at the net-
sentiment analysis of these five clusters of tweets is also a bell- work closely reveals that these tend to be key hubs in their
curve shape. Risk-related tweets appear to be more negative than communities of nodes. For example, @logisticsmatter is a hub
the other groups of tweets. node in a community of users interested in logistics, and @Guar-
Table 3 shows exemplar negative and positive tweets from each dianSustBiz and @schwild are in communication paths between
group. For instance, a negative risk-related tweet (“With new users discussing CSR and related topics. While centrality is low,
#supplychain controls in place IKEA meatballs back after horse- supply chain companies, including logistic providers (e.g., @UPS),
meat scare http://t.co/YKvYT1ov1H#ris”) is referring to an article manufacturers (e.g., @unilever), and retailers (e.g., @walmart) are
published in the business section of the Chicago Tribune, which present in the dataset.
describes the incident of horsemeat being found in the meatballs Community analysis (network-level metrics): The community
from Ikea's main supplier, Familjen Dafgård in Sweden. analysis shows a very low graph density (0.001), indicating that
the entire #supplychain network is sparsely distributed (not
cohesive), as shown in Fig. 5. This may be an indication that there
4.3. Network analytics (NA) are many dispersed groups in the network. We used a community
detection method (Blondel et al., 2008), embedded in Gephi, an
Topological analysis: A large social network was constructed open source graph software tool. This method reveals over 400
including 5447 nodes and 9238 edges. In this network graph, the communities in the #supplychain network. The majority of these
nodes are those users who either sent out or received @reply. The communities are quite small-scale. There are four large commu-
edges are the relationships between those users through @reply. nities, each of which represents over 5% of the total nodes, about
The path length of most nodes is between 3 and 8. The network corporate social responsibility, sustainability, manufacturing, and
diameter is found to be 19, which is the longest path between two SCM software. The largest community is a community of CSR and
nodes in the network. The typology (Fig. 5) indicates that it is a sustainability, which represents 8% of the total users. The second

Fig. 3. Sentiment analysis at the entire dataset level (the numbers on x-axis Fig. 4. Sentiment analysis of the clustered tweets (CSR, IT, Manufacturing,
indicate the degree of either positive or negative sensitivity). Logistics, Risk).

Table 2
Exemplar tweets with relatively strong sentiment.

Exemplar tweets Sentiment

This development is pretty exciting for me #supplychain #sustainability 3 (positive)

Samsung assures excellent working conditions in China. 〈http://t.co/9VcBIL1XAj #workingconditions #supplychain #responsiblesourcing #rtw〉 3 (positive)
Loving the feedback that I am getting on this report today! #supplychain 〈http://t.co/7DJaJFFl〉 3 (positive)
As parked 787 s multiply? Boeing cash drain worries grow 〈http://t.co/up6V9sBYbi #supplychain〉  3 (negative)
CHaINA Magazine: Apple Finds Child Labor Abuses in Its Supply Chain – As the controversy regarding. 〈http://t.co/kt6MMXvng #SupplyChain〉  3 (negative)
Walmart US Chief Says Sales are Suffering Because of Restocking Issues 〈http://t.co/xhpK4pYy3v #supplychain〉  3 (negative)
 B.(K.) Chae / Int. J. Production Economics 165 (2015) 247–259 253

Table 3
Exemplar tweets from five clusters of tweets.

Exemplar tweets Sentiment

CSR Investigation of #Apple #SupplyChain by #SACOM reveals continued #labor abuse throughout despite Apple's claims. 〈http://t.co/ Negative
Cm0VYeTT94 #CSR〉
Loving this Green Supply Chain infographic 〈http://t.co/M0spSUs6WG #supplychain #CSR #infographic @SCInsightsLLC〉 Positive
IT Supply Chain Disruption a Major Threat to Business from @forbes 〈http://t.co/raX48u4Zcw #supplychain #IT #analytics〉 Negative
How @Sony strengthened its supply chain and added value 〈http://t.co/B0DTiWPAnm #GrnBz via @GreenBiz #IT #supplychain〉 Positive
Logistics Could I expect to receive my parcel today? I doubt it. Another terrible experience from @Parcel2Go #expensive #supplychain #logistics Negative
Great day @ #DalhousieUniversity! Excellent case study presentations by 16 teams. Join Careers in Motion @JDIcareers #logistics Positive
#supplychain
Manufacturing Central China province suffers post-Spring Festival #labor shortage #manufacturing #SupplyChain #reshoring 〈http://t.co/fX09sxmlxN〉 Negative
Excellent insights: Predictions from #SupplyChain Gurus for 2103. 〈http://t.co/MQFZSgTI (See #manufacturing at end) Positive
Risk With new #supplychain controls in place IKEA meatballs back after horsemeat scare 〈http://t.co/YKvYT1ov1H #risk〉 Negative
Thanks! RT @riskleadership @GenexLogistics @Kinaxis: #SupplyChain Excellence in Words and Numbers by @indexgirl 〈http://t.co/ Positive
9gV2cXD2〉

5.1. What are the characteristics of supply chain tweets? Are there
any patterns of communication and information diffusion?

There are different reasons for individual and organizational

users' tweeting. Twitter users disseminate information using URLs,
chat about daily routines, engage in conversation with @reply,
and report news about events and incidents (Java et al., 2007). In a
random sample of tweets, the rate of retweet is just 3%, and 22%
of tweets contain a URL (Boyd et al., 2010). In contrast, 89% of
#supplychain tweets include a URL and 28% of them are retweets.
Also, the other 16% of tweets contain @user, meaning these tweets
are mentioning one or more other users. Overall, this indicates
that supply chain tweets appear to be more conversational and
Fig. 5. The topology of “the supply chain sphere” in Twitter. engaging than random samples of tweets. Supply chain profes-
sionals and organizations use Twitter for reporting news and for
information sharing. About 12% of randomly selected tweets
contain a conversation (@reply) (Java et al., 2007). This rate is
slightly lower than that of #supplychain tweets (14%), indicating
Table 4 that #supplychain tweets appears to be more conversational than
High in-degree users.
public tweets.
User In-degree Out-degree Degree The rate of hashtags in #supplychain tweets also draws atten-
tion. Users include hashtags to indicate that their tweets are
Ilmagazine 221 57 278 topical, so other users with similar interests can follow. While it
logisticsmatter 142 6 148 is not clear how many randomly sampled tweets would contain at
industryweek 127 0 127
deloitte 108 6 114
least one hashtag, we expect a relatively low percentage. Only
mfgexecutive 103 1 104 13–23% of tweets by software engineers, for instance, contain at
CSRwire 95 26 121 least one hashtag (Bougie et al., 2011). In constrast, a large portion
Lcecere 80 57 137 (73%) of #supplychain tweets appear to contain two or more
GuardianSustBiz 79 16 95
hashtags. On average, almost 3 hashtags appear in tweets. In addi-
ajsweatt 79 25 104
toyotaequipment 76 111 187 tion, there is a surprisingly large number of hashtags (3839) found
logisticsviewpt 75 9 84 in #supplychain tweets. Among those topics, logistics, manufac-
SustainBrands 72 8 80 turing, sustainability, corporate social responsibility (CSR), risk
Kinaxis 59 45 104 management, and IT account for a large portion of the total
sdcexec 59 0 59
SCDigest 54 0 54
number of tweets.
We found that two factors play a key role in determining infor-
mation diffusion of supply chain tweets. First, tweets about timely
issues and challenges tend to be more widely diffused than others.
largest community is focusing on manufacturing and distribution. This is evident in that a disproportionately large number of tweets
5% of the total users belong to this community. The SCM software containing such hashtags as #manufacturing, #sourcing, #reshor-
community includes another 5% of users. ing, #procurement, #retail, #sustainability, #CSR, #risk, #BigData,
#SocialMedia, and #food are widely diffused through retweeting. In
contrast, job-related tweets are rarely retweeted. This shows that
tweets concerning new trends (e.g., #BigData) and issues (#risk,
5. Discussion #sustainability, #manufacturing) in SCM are propagated widely.
Second, the number of hashtags is positively associated with the
In this section, we discuss our findings in light of those four degree of diffusion through retweets. We further examined those
questions raised in the introduction. This discussion will also serve top retweeted tweets. The result shows that those tweets widely
as the basis for Section 6. diffused through retweets tend to contain six hashtags on average,
254 B.(K.) Chae / Int. J. Production Economics 165 (2015) 247–259

which is twice that of other less popular tweets. Our findings are logistics companies. A second group includes SCM practitioners.
partially alinged with studies of Twitter data in other fields (Stieglitz Another active group includes SCM magazines, IT service companies
and Dang-Xuan, 2013; Suh et al., 2010; Yuan et al., 2012). These (e.g., @Analytiqa, @Kinaxis), and news service companies (or aggre-
studies also suggested other potential factors for information gators) (e.g., @logisticsmatter, @CSRwire, and @supplychain_fm). Not
diffusion, such as the number of followers and following (Suh all active users are highly visible. Visible users receive many @reply
et al., 2010), and the degree of tweet sentiment (Stieglitz and (or mentions) and their tweets are retweeted. Highly visible users
Dang-Xuan, 2013). come from diverse areas (e.g., CSR, manufacturing) and roles (e.g.,
SCM experts, online magazines, professional organizations, and IT
5.2. What supply chain topics or contents are shared in Twitter? companies). Some major logistic providers (e.g., UPS), manufacturers
Are there any prevalent topics or contents? (e.g., Unilever, Sony), and retailers (e.g., Walmart) are visible, but
most of them are not very active in #supplychain tweets. This seems
Supply chain topics are very diverse, as evidenced by the to be because these large carriers' and manufacturers' Twitter
number of hashtags (3839) in the data. Most popular hashtags accounts tend to be used to send promotional messages to their
appear to be similar to topic areas covered in academic SCM followers, as part of a brand management strategy, rather than
journals. They include #manufacturing, #logistics, #sustainability, engage with their business partners and clients in SCM contexts.
#procurement, #csr, #3PL, #sourcing, #IT, #distribution, #transpor- The analysis also shows that a relatively small percentage of
tation, #risk, #healthcare, #ERP, #riskmanagement, #innovation, users account for a large portion of tweets. For example, an
#purchasing, #lean, and #inventory. A close examination also analysis of over 11 million tweets found that only 5% of users
indicates that topics of interest in tweets are broader than what account for 75% of tweets (Cheng et al., 2009). A random sample of
we generally discuss about SCM in academic press. They include 300,000 tweets also shows that 10% of users account for over 90%
#BigData, #SocialMedia, #careers, #equity, #CRM, #SaaS, #human- of tweets (Heil and Piskorski, 2009). This pattern appears very
rights, #horsemeat, #fashion, #RFID, #fairtrade, #ClimateChange, strong in #supplychain tweets. Only 1% (45 users) and next top 9%
#Android, #ethics, #analytics, and #humantrafficking. (395 users) account for almost 40% and 34% of tweets, respectively.
One noticeable hashtag is #humanrights. On April 24, 2013, there This is more evident in the ratio of users and original tweets. The
was a deadly garment factory accident4 in Bangladesh, which has led top 10% of users account for 52% and 33% of original tweets.
to a great deal of discussion of SCM, with topics such as #safety, However, the other 90% of users account for a large portion (50%)
#consumergoods, #csr, #sustainability, #ethics, and #humanrights in of retweets, indicating this group of users tend to be followers of
social media, particularly in Twitter, and the academic press. Our data those highly visible users in the network.
collection was finished before this accident. Despite this, our data
also contain numerous tweets about #humanrights and #supply-
chain. First, some of these tweets remind Twitter users of another 5.4. What are the sentiments of supply chain tweets? What types of
previous garment factory accident, also in Bangladesh, in November supply chain tweets tend to contain sentiment?
2012 (“#Bangladesh factory fires – the hidden dangers of subcon-
tracting. http://t.co/taEDvIwZ #supplychain #labour”). Some also #supplychain tweets contain relatively low sentiment. While
discuss supply chain strategy at such events (“What is #Walmart's this finding was initially surprising, the findings from descriptive
long-term sourcing strategy in Bangladesh? http://t.co/Wuv1I4TjkR analytics (DA) and content analytics (CA) help explain why. Unlike
#supplychain #humanrights”). Others warn of another accident in those tweets, either for or against certain political candidates and
Bangladesh and elsewhere (“Bangladesh fires reinforce pressing need movies, most #supplychain tweets were about events, SCM news
for greater #supplychain transparency http://t.co/HYdCfkIoiu”). and reports, jobs, and advertisements. Thus, those tweets tend to
The analysis of words shows that wording in the entire tweets is contain weak sentiment. Another explanation would be that there
similar to that of hashtags. However, the word analysis of clustered was no major SCM-related incident during the period of data
documents shows that there are distinctive words used in clustered collection. Major events, such as the Bangladesh garment factory
tweets. Logistics is the most popular word in tweets. The tweets collapse, draw emotional, largely negative, tweets. Nevertheless,
about logistics include such words as careers, software, trucking, some tweets were found to be carrying strong disappointments
freight, transportation, and shippers. In contrast, the tweets about about a company's delivery service, sales performance, and ethical
sustainability include such words as csr, food, csrwire, improving, (or environmental) standards, and risk and disruption in the
and green. Popular URL domains seem to coincide with the analysis supply chain.
of hashtags and words. They are related to careers and jobs, SCM Exemplar tweets are “Nine Retailers with the Worst Customer
news outlets, and SCM IT companies. The diversity of supply chain Service - 24/7 Wall St. http://t.co/BcBZ63meXs #business #supply-
topics is also evident in the findings of over 400 communities from chain #manufacturing #logistics”, “Investigation of #Apple #Sup-
network analytics. Noticeably, corporate social responsibility, sus- plyChain by #SACOM reveals continued #labor abuse throughout
tainability, manufacturing, SCM software, and logistics represent despite Apple's claims. http://t.co/Cm0VYeTT94 #CSR”, and “Central
some of the largest topical communities. Many small communities China province suffers post-Spring Festival #labor shortage #man-
represent groups of other SCM topics. ufacturing #SupplyChain #reshoring http://t.co/fX09sxmlxN”. Posi-
tive sentiment was found in promotional tweets (“How @Sony
strengthened its supply chain and added value http://t.co/B0DTiW-
5.3. What are the characteristics of those Twitter users who are PAnm #GrnBz via @GreenBiz #IT #supplychain”; “Great day @
discussing supply chain topics? #DalhousieUniversity! Excellent case study presentations by 16
teams. Join Careers in Motion @JDIcareers #logistics #supplychain”).
Our analysis reveals important information regarding who is Among the tweets, those related to supply chain risk, manufac-
active in Twitter. Some of the most active users are job or career turing, and CSR are considered more sentimental than others (IT,
services. Nearly all tweets from this group of users are original logistics). Noticeably, risk-related tweets tend to carry far more
tweets, which are job announcements for supply chain and negative sentiment than those of other topics. There appear to be
disproportionately more negative tweets in the topic of manufactur-
4
This incident occurred on April 24, 2013, which is right after our formal data ing. Exemplar tweets are “Quality of #manufacturing and #supply-
collection was completed. chain #education is a growing concern http://t.co/jfswNukNH”,
 B.(K.) Chae / Int. J. Production Economics 165 (2015) 247–259 255

Table 5
Twitter's potential role for supply chain professionals and companies.

Examples

Professional use
Learning  Following experts on the topics such as supply chain analytics
 Searching topics (or keywords)

Promoting  Tweeting/retweeting useful information and opinions on timely topics and issues

Networking  Using such Twitter features as @reply and follower/following

Organizational use
Stakeholder engagement  Twitter as a communication platform
 Spreading positive images as socially responsible and successful companies
 Reaching out to a large public with success story through tweets and retweets

Hiring  Twitter as a hiring tool

 Tweeting job openings and descriptions
 Mining Twitter user timeline and identifying talented professionals using descriptive, content, and network
analytics

Demand shaping and sales  Twitter as a sales channel

 Tweeting production information and encouraging followers to retweet
 Creating customer demands by tweeting coupons and discounts

Market sensing and new product/service  Twitter as a market sensor

development  Extracting demand signals from customers and markets using sentiment analysis
 Getting customers' ideas and feedback about products, quality, and service

Risk management  Twitter as an event monitoring and collaboration tool

 Sensing supply-chain related events, disruptions, and other news in real time
 Broadcasting supply chain events and risks in real time to supply chain partners and helping collaboration

“Supply chain risk on the rise http://t.co/cMeWWotx #manufactur- Compared to traditional media (e.g., web pages, books, and confer-
ing #SupplyChain” and “#FastFashion retailer #Zara Faces #Slave- ences), the key characteristic of microblogging is that users do not
Labor Allegations. Again http://t.co/nTJWxjzus3 via @ecouterre #CSR need to be proactive in acquiring information. Twitter makes a large
#manufacturing #supplychain”. “Bad news tends to travel fast” volume of information delivery/search and knowledge acquisition
(Naveed et al., 2011). This may explain the proportionately high faster and more convenient, for example, by following experts on the
diffusion rate of those tweets with #manufacturing and #risk. topics and/or simply searching topics (or keywords). For example, the
simple search5 using “supply chain analytics” (c.f., Chae et al., 2014)
at twitter.com provides opportunities for learning such a new topic
6. Implications: the use and potential role of Twitter for through browsing relevant news, videos, and photos, and finding
supply chain practice and research other professionals and organizations in this emerging area.
In addition, promoting one's expertise and networking with
Derived from the proposed analytical framework in Section 3, others is important in the professional world. Academics and
its application in Section 4, and the research findings in Section 5, journals use Twitter for promotion of their works and contents
below we present further implications of Twitter for professional (Holmberg and Thelwall, 2014; Nason et al., 2014). Likewise, Twitter
use, organizational use, and academic research. A summary is can be an effective tool for supply chain professionals to pro-
presented in Table 5. mote themselves, for example, through tweets/retweets of useful
information and opinions on timely topics and issues. Examples
6.1. Professional use of Twitter are “Only 36% of #supplychain leaders use predictive modeling
to forecast supplier risk and worst case scenarios http://t.co/
Professionals in different fields have used Twitter for a variety mB3cxyuACU”; “I shared this with my workshop friends from
of reasons (Conway et al., 2013; Vis, 2013). The findings of today‚ so I wanted to share with you too. #supplychain http://t.co/
analyzing supply chain-related tweets indicate that they are more 9xqvZUa2MO”. Unlike those traditional media that require signifi-
information-focused and conversational than regular tweets. This cant time and effort to prepare contents for self-promotion and
is evident from the inclusion of URLs and the degree of informa- opinion sharing, in Twitter this can be accomplished by a short
tion diffusion through retweeting. These findings suggest that message including hashtags, URLs, videos, and/or photos. This
Twitter can be the platform for learning, promoting and network- “brevity” of microblogging explains the growth of tweets among
ing, and some supply chain professionals are already taking advan- professionals over the past several years. Also, such Twitter features
tage of Twitter for such activities. as @reply and follower/following enable supply chain professionals
First, learning is critical for the success of all professionals, but this to engage with other professionals and clients in real time
is particularly true for supply chain professionals in a dynamic and
fast changing business landscape. For them, Twitter is an important
source of finding the latest news and events, and acquiring new 5
https://twitter.com/search?q=supply%20chain%
knowledge about a broad range of supply chain topics and issues. 20analytics&src=typd&lang=en.
256 B.(K.) Chae / Int. J. Production Economics 165 (2015) 247–259

(“@walmart is aiming to bring #women into their #supplychain‚ is does (“Interested in a #career in Supply Chain with #Unilever
it working? http://t.co/eG4LDK2Eec #IWD #womensday”; “Thx for Learn more @ http://t.co/s8slqtNoNk #CPG #Jobs #Diversity
sharing! MT @SCBrain: Great Resource! #SocialMedia in the #Sup- #SupplyChain”).
plyChain: http://t.co/UYKaCTPT87”). The role of competent professionals is critical for companies to
Supply chain professionals can take advantage of those analy- stay competitive (Cottrill, 2010). As firms' supply chains are becoming
tical techniques and metrics in the proposed framework for global and more complex, hiring such people is a top priority for
learning, promoting, and networking. For example, the network companies (Hohenstein et al., 2014). Social media emerges as a tool
analytical techniques and metrics can be utilized to figure out their to recruit supply chain professionals (Fisher et al., 2014). Twitter can
networks and communication patterns for effective networking be used for both passive and active recruiting strategies. Passive
and promoting. Also, the descriptive analytics and metrics (e.g., recruiting is like tweeting job descriptions, as many companies in our
number of replies and retweets) can indicate the popularity and dataset do. In addition, companies actively identify potential candi-
reputation of professionals. Content analytics, especially sentiment dates or experts through analyzing user profile information and user
analysis of the tweets mentioning them and the retweets, can timelines. The user profile of supply chain professionals includes
enable professionals to measure how their tweets and opinions their job experiences and other social media accounts and/or web
are perceived by other professionals and clients. pages for detailed profile information. Mining user timeline (e.g.,
tweets, retweets, following, photos/videos, favorites), using descrip-
6.2. Organizational use of Twitter tive, content and network analytics, can offer valuable information,
such as reputation, expertise, and professional network.
The research findings from #supplychain tweets show that In addition to the currently popular uses of Twitter described
Twitter is currently widely used for stakeholder engagement, above, there are other potential uses of Twitter by supply chain
company promotion, and hiring. companies. They are demand shaping and sales, market sensing
Communication and stakeholder engagement: The data and and new product/service development, and supply chain risk
content analytics show that, for some logistics providers, manufac- management, among others.
turers, and retailers, the use of Twitter is strong in the area of Demand shaping and sales: Twitter can be a channel for sales. In
stakeholder engagement. Especially, the focus has been on spread- 2009, Dell announced it had made over $3 million in sales, since
ing their positive images as socially responsible and successful 2007, from those who visited their website directly from Twitter,
companies. For example, Unilever, a global consumer goods com- during the time most companies had little understanding of using
pany ranked in Gartner's list of top supply chain companies, shows Twitter and other social media platforms (Miller, 2009). Twitter is
much interest in using Twitter for reaching out to a large public suitable for reaching out to a large number of potential consumers, as
with their success story (“RT @2degreesnetwork: @Unilever is going companies tweet product information, coupons, and discounts, and
for the top spot in #supplychain mgmt at the #2degreesAwards this information appears in the followers' timelines. For example,
http://t.co/vHWODYXk1k RT @CSRwire”; “We're committed 2our Dell Outlet and Starbucks, among those recognized in the Gartner's
#supplychain but we'll never lose sight of the value of embedding list of top supply chain companies, have about 1,430,000 and
#sustainability 4our #brands http://t.co/OYr9M1jPVH”). Also, other 6,720,000 followers respectively. Some of these followers are likely
companies such as DHL seem to use Twitter to increase public to retweet coupons and discount information, which affect the
awareness of their brand and business activities (e.g., “RT @Analy- purchasing decisions of their followers. Companies can even encou-
tiqa: DHL Supply Chain invests in Indonesia; increases in fleet‚ rage retweeting to their followers through rewards (e.g., more
warehouse footprint‚ and staff #Logistics #SupplyChain”). discounts). The result is the increase in sales, as shown in some
This use of Twitter is expected since microblogging is about recent studies of tweets and corporate sales (Gong et al., 2014).
communication, and such features (e.g., brevity, fast updates, @reply, This Twitter effect on sales is due to what is broadly named
hashtag, retweet) of Twitter offer companies great opportunities to “electronic word of mouth”. Word of mouth is popularly recog-
engage with a large number of stakeholders more quickly and nized as a powerful mechanism, wherein consumers' decisions or
conveniently than via traditional media. Communicating supply opinions are spread to others and affect their purchasing decisions.
chain initiatives (e.g., sustainability) and positive news (e.g., innova- Twitter increasingly plays the role of the platform for electronic
tion) with stakeholders is considered among the strategic uses of word of mouth (Jansen and Zhang, 2009). The SCM field has long
Twitter, since a message can be “viral” in Twitter or, at minimum, is emphasized demand shaping as a best practice adopted by top
viewed by followers. According to a report (Telegraph, 2012), the supply chain organizations to balance supply and demand.
average number of followers per Twitter account is over 200. Demand shaping is “a demand-driven, customer centric approach
However, this number is very high for large corporations. For to supply chain planning and execution … influencing customer's
example, Unilever and UPS have 67,500 and 110,971 followers, demand towards products that the firm can supply easily and
respectively. This means Unilever's tweet is broadcast to a minimum profitably” (Dietrich et al., 2012, p. 1). Companies can greatly
of 67,500 followers, some of whom retweet. Currently, there is a increase their demand shaping capacity through strategically
severe imbalance between the number of following (about 4500 for formulated tweets, which aim to create consumer demands. Using
both Unilever and UPS) and the number of followers, indicating the descriptive analytics (e.g., number of retweets), they can measure
communication is likely unidirectional, rather than dialogic. Supply the effectiveness of their tweets for demand shaping and sales.
chain organizations should consider following more stakeholders for Market sensing and new product/service development: “Could I
dialogic communication and engagement. expect to receive my parcel today? I doubt it. Another terrible
Hiring: As noted earlier in the research findings, another strong experience from @Parcel2Go #expensive #supplychain #logistics.”
area of Twitter use is hiring. The data contain a large number of In addition to its role as a sales channel, Twitter can be an effective
tweets about supply chain job opportunities, ranging from truck- “sensor”, providing signals from customers and markets. The
ing to engineering. For example, General Mills, a food product analysis of #supplychain tweets clearly indicates that tweets con-
manufacturer and Fortune 500 corporation, tweeted about a tain emotion, opinions, and preferences. This means Twitter has the
specific job “#Hiring: Plant Environmental Health and Safety potential to be a sensor platform. Market sensing is considered a
Manager at our #Chanhassen‚ MN plant: http://t.co/3qMN2Y4r dynamic capability significantly impacting the performance of
#supplychain #engineering #job”. On the other hand, others try to organizations (Teece, 2007) by increasing traceability and tracking
draw potential job candidates to their career page, as Unilever and understanding market demands (Beske et al., 2014). Twitter
 B.(K.) Chae / Int. J. Production Economics 165 (2015) 247–259 257

data has been the source of successfully predicting customers' to the contents of tweets deserve further discussion. Almost 4 of
adoption of different types of products and services (Arias 5 tweets include two or more hashtags and over 3800 hashtags are
et al., 2014; Hennig-Thurau et al., 2014). The big data from Twitter found in the Twitter data collected over a two month period. The
and sentiment analysis (content analytics) can greatly help compa- variety of topics discussed with hashtag #supplychain is astonish-
nies be effective in learning the public's perception of products (e.g., ing. This finding indicates that supply chain is relevant to a very
quality) and services (e.g., delivery) (O’Leary, 2011). This market- large number of social practices (e.g., sustainability), and business
sensing data and analysis are critical for companies' sales and and IT applications (e.g., big data). This variety in supply chain is
operations planning, which synchronizes market demand and also evident in other findings (e.g., a large and heterogeneous
supply capabilities (Tuomikangas and Kaipia, 2014), and can com- network topology, over 400 communities) from network analytics.
plement extant demand forecasting methods (e.g., smoothing) and Thus, these findings imply that the field of supply chain is shown to
processes used in practice (Ferbar et al., 2009). be as diverse as the number of hashtags and communities. This
In this vein, more companies are considering Twitter for getting similar finding also appears in the academic research of “the
customers' ideas and feedback about products, quality, and serv- intellectual structure of supply chain management discipline”
ices. For example, Dell has used Twitter to respond to customer (Giannakis, 2012). The citation-based research indicates that SCM
complaints and actively seek customer suggestions, as shown in the is fundamentally interdisciplinary and there is much more diversion
development of Dell Mini 10 based on Twitter users' complaints on than cohesion in the field.
Dell Mini 9 (Miller, 2009). Regarding Twitter use in this area, a Dell Second, among those many topics and interests, our findings
personnel notes “It's a great way to fix customer problems and hear from the analysis of words, hashtags, and communities show that
what customers have to say, it's a great feedback forum … how can sustainability and its relevant topics (e.g., corporate social respon-
you miss?” (Miller, 2009). Customer knowledge is critical for sibility, green management) have moved into the center of supply
successful product/service development (Chae, 2012). Starbucks chain-related discussions in Twitter. This growth of sustainability-
uses Twitter and other social media by redefining the role of related discussions in social media is no different than academic
consumers from passive recipients to co-creators in product/service research. Despite its relatively short history in academic research,
innovation process (Chua and Banerjee, 2013). the topic of sustainable supply chain management has become the
Risk management: “Helicopter hovering above Abbottabad at mainstream research area (Chen et al., 2014; Giannakis, 2012;
1 AM (is a rare event).” This is a tweet by an IT consultant working Seuring and Müller, 2008). Research communities can identify
late at night and unknowingly reporting the raid of terrorist emerging supply chain topics and issues important to practitioners
Osama bin Laden hours before any news channels broke the news. and companies from Twitter data.
A key characteristic of Twitter is the real time broadcasting of Finally, the research design and findings show the possibility of
diverse events and accidents. Over 270 million users worldwide using Twitter data for supply chain research. Researchers in diverse
are potentially reporters or sensors (Sakaki et al., 2010) of events, academic fields have used Twitter data for many important research
such as disasters (e.g., Hurricane Sandy, Japan earthquakes), crises questions. In contrast, there are few studies using Twitter (and
(e.g., Boston Marathon Bombing), and critical social changes (e.g., social media platforms) in the SCM field. We believe that Twitter
Arab Spring) (Hughes and Palen, 2009). through API can offer SCM researchers (and industry practitioners)
Supply chain risk management (SCRM) has emerged as a top the opportunity to access not just open public data, but also “big
priority for companies and a key stage of SCRM is detecting potential data”, which are significant in terms of scale, size, and speed. Social
disruption (Chopra and Sodhi, 2014). Twitter, as the sensing and media in general and Twitter in particular, can serve as new data
broadcasting platform, offers great opportunities for SCRM since sources and collection methods, which could complement existing
tweets broadcast a variety of supply chain-related events, disrup- data sources and collection methods for SCM research and practice
tions, and other news in real time. Examples are “Thailand braces for (c.f., Calantone and Vickery, 2010).
blackouts #supplychain http://t.co/WNmMoBM1Vr”, “Honda recalls
250‚000 vehicles globally http://t.co/f8E2PSvPxU #risk #riskmanage-
ment #Insurance #marketing #supplychain” and “Strong earthquake 7. Limitations and future research
sways buildings throughout Taiwan http://t.co/3i2SK7tw3L via
@CTVNews #supplychain.” Some tweets (e.g., “holy cow, I AM in It is imperative for supply chain professionals and organiza-
Japan,… earthquake, earthquake right now!”) may be created before tions (and even researchers) to effectively use social media plat-
such an earthquake is registered with the US Geographical Survey forms and social media data for their operations. However, the
(Ostrow, 2009). SCM field has been relatively slow in identifying the potential roles
To respond to these potential risks, a supply chain needs to facilitate and uses of social media and social media data. Academic research
communication among supply chain personnel. Microblogging is and reported industry cases are rare on this important topic. This
suitable for instant communications and broadcasting them to a large paper has responded to the current situation by proposing an
population. “Dorsey's [co-founder of Twitter] genealogy of Twitter analytical framework for Twitter data, highlighting a broad range
refers to communication systems for bicycle messengers, truck cour- of intelligence (e.g., popular topics, communication patterns, tweet
iers, emergency services, ambulances, fire trucks and police” (Rogers, sentiments), the extant use of Twitter in SCM contexts using the
2013, p. 1). There are a variety of supply chain-related events (e.g., road proposed framework, and finally developing further insights into
accidents, flood), which can cause supply chain risks and disruption innovative uses of Twitter for supply chain practice and research.
(Tang and Nurmaya Musa, 2011). Twitter has the potential to be a This study has limitations, particularly related to data collection.
communication system for logistics and transportation personnel to The first is the relatively short duration of data collection, which
broadcast such events and risks in real time to the partners in their lasted a little over two months. Data collection for a longer period is
supply chain. This can help with supply chain risk detection and expected to develop a more complete picture of the use of Twitter
disruption recovery (Kumar and Havey, 2013; O’Leary, 2011). for SCM. Related to this is the way the data were collected. As noted
earlier, we chose hashtag #supplychain to search and collect Twitter
6.3. Supply chain research data. Another, perhaps better, approach would be using multiple
keywords (e.g., supply chain) and/or hashtags (e.g., #manufacturing,
Aside from those implications for practical applications, impor- #logistics) in data collection. This approach would enable research
tant research implications can be derived. First, our findings related using a large quantity of supply chain-related Twitter data. Also, a
258 B.(K.) Chae / Int. J. Production Economics 165 (2015) 247–259

large volume of tweets, while they are useful for companies' supply Cheng, A., Evans, M., Singh, H., 2009. Inside Twitter: An In-Depth Look Inside the
chain decisions, may not contain such keywords as supplychain, Twitter World. 〈http://www.sysomos.com/docs/Inside-Twitter-BySysomos.pdf〉
(accessed 20.07.14).
manufacturing, logistics, risks, etc. A solution to remedy these cases Chopra, S., Sodhi, M., 2014. Reducing the risk of supply chain disruptions. MIT Sloan
is to acquire Twitter data using a large list of keywords. Manag. Rev. 55, 72–80.
We believe there is much need for enhancing our understand- Chua, A.Y.K., Banerjee, S., 2013. Customer knowledge management via social media:
the case of Starbucks. J. Knowl. Manag. 17, 237–249.
ing of social media and social media data in supply chain contexts. Conway, B.A., Kenski, K., Wang, D., 2013. Twitter use by presidential primary
At least two broad streams of academic research and practical candidates during the 2012 campaign. Am. Behav. Sci. 57, 1596–1610.
applications should be followed. One area is developing detailed, Cottrill, K., 2010. Are You Prepared for the Supply Chain Talent Crisis? MIT CTL
White Paper 〈http://www.distributiongroup.com/articles/0211mit.pdf〉 Fall,
practical guidelines, which can help companies in designing
pp. 1–11.
industry applications, using Twitter and other social media plat- Daniel, Z., Hsinchun, C., Lusch, R., Shu-Hsing, L., 2010. Social media analytics and
forms, for diverse supply chain activities, including new product intelligence. IEEE Intell. Syst. 25, 13–16.
development, stakeholder engagement, supply chain risk manage- Dataminr, 2014. Dataminr's Event Detection Technology. 〈https://www.dataminr.
com/technology/〉 (accessed 01.08.13).
ment, and market sensing. Research programs partnering with Dickey, M., 2014. Twitter Gears Up To Launch A TweetDeck On Steroids For
innovative organizations can play important roles in this area. In Journalists. Business Insider 〈http://www.businessinsider.com/twitter-and-data
this line, supply chain researchers should first study the innovative minr-2014-1〉 (accessed 18.03.14).
Dietrich, B., Ettl, M., Lederman, R., Petrik, M., 2012. Optimizing the end-to-end value
applications of social media in other disciplines. The other area is chain through demand shaping and advanced customer analytics. In: Karimi, I.,
studying the impact of social media (and big data) investment Srinivasan, R. (Eds.), Proceedings of the 11th International Symposium on
(e.g., technologies, data scientists, and data repositories) on supply Process Systems Engineering. Elsevier, Singapore.
Dodds, P., Harris, K., Kloumann, I., Bliss, C., Danforth, C., 2011. Temporal patterns of
chain performance. This type of research is urgently needed, as happiness and information in a global social network: hedonometrics and
there is increasing attention to and interest in big data and social Twitter. PLoS One 3, 1–26.
media from industries and academia. Fan, W., Gordon, M.D., 2014. The power of social media analytics. Commun. ACM 57,
74–81.
Feldman, R., 2013. Techniques and applications for sentiment analysis. Commun.
ACM 56, 82–89.
Ferbar, L., Čreslovnik, D., Mojškerc, B., Rajgelj, M., 2009. Demand forecasting
References methods in a supply chain: smoothing and denoising. Int. J. Prod. Econ. 118,
49–54.
Aral, S., Dellarocas, C., Godes, D., 2013. Social media and business transformation: a Fisher, R., McPhail, R., You, E., Ash, M., Richey, G., 2014. Using social media to recruit
framework for research. Inf. Syst. Res. 24, 3–13. global supply chain managers. Int. J. Phys. Distrib. Logist. Manag. 44, 635–645.
Arias, M., Arratia, A., Xuriguera, R., 2014. Forecasting with Twitter data. ACM Trans. Galaskiewicz, J., 2011. Studying supply chains from a social network perspective.
Intell. Syst. Technol. 5, 1–24. J. Supply Chain Manag. 47, 4–8.
Bennett, S., 2013. Twitter Was The Fastest-Growing Social Network in 2012, Gayo-Avello, D., 2012. A Meta-analysis of State-of-the-art Electoral Prediction From
Says Study [STATS]. 〈http://www.mediabistro.com/alltwitter/social-networks- Twitter Data. arXiv:1206.5851v1 [cs.SI].
growth-2012_b35076〉 (accessed 01.09.13). Georgi, C., Darkow, I.-L., Kotzab, H., 2010. The intellectual foundation of the Journal
Beske, P., Land, A., Seuring, S., 2014. Sustainable supply chain management of Business Logistics and its evolution between 1978 and 2007. J. Bus. Logist. 31,
practices and dynamic capabilities in the food industry: a critical analysis of 63–109.
the literature. Int. J. Prod. Econ. 152, 131–143. Giannakis, M., 2012. The intellectual structure of the supply chain management
Bird, S., Klein, E., Loper, E., 2009. Natural Language Processing with Python. O’Reilly discipline: a citation and social network analysis. J. Enterp. Inf. Manag. 25,
Media, Cambridge. 136–169.
Blondel, V., Guillaume, J., Lambiotte, R., Lefebvre, E., 2008. Fast unfolding of Gong, S., Zhang, J., Zhao, P., Jiang, X., 2014. Tweets and Sales. Available at SSRN:
communities in large networks. J. Stat. Mech.: Theory Exp. 10, 10008. 〈http://ssrn.com/abstract=2461370〉 or doi: 10.2139/ssrn.2461370 (01.07.14).
Bollen, J., Mao, H., Zeng, X., 2011. Twitter mood predicts the stock market. Harris, D., 2014. 3 Lessons in Big Data from the Ford Motor Company.
J. Comput. Sci. 2, 1–8. Hazen, B.T., Boone, C.A., Ezell, J.D., Jones-Farmer, L.A., 2014. Data quality for data
Borgatti, S., Lin, X., 2009. On social network analysis in a supply chain context. science, predictive analytics, and big data in supply chain management: an
J. Supply Chain Manag. 45, 5–22. introduction to the problem and suggestions for research and applications. Int.
Bougie, G., Starke, J., Storey, M., German, D., 2011. Towards understanding Twitter J. Prod. Econ. 154, 72–80.
use in software engineering: preliminary findings, ongoing challenges and Heil, B., Piskorski, M., 2009. New Twitter Research: Men Follow Men and Nobody
future questions. In: Proceedings of the 2nd International Workshop on Web Tweets. HBR Blog Network 〈http://blogs.hbr.org/2009/06/new-twitter-re
2.0 for Software Engineering. ACM, pp. 31–36. search-men-follo/〉 (accessed 04.01.13).
Boyd, D., Golder, S., Lotan, G., 2010. Tweet, tweet, retweet: conversational aspects of Hennig-Thurau, T., Wiertz, C., Feldhaus, F., 2014. Does Twitter matter? The impact of
retweeting on Twitter. In: Proceedings of the 43rd IEEE Hawaii International microblogging word of mouth on consumers' adoption of new movies. J. Acad.
Conference on system Sciences. Kauai, HI, pp. 1–10. Mark. Sci., 1–20. http://dx.doi.org/10.1007/s11747-014-0388-3.
Bruns, A., Burgess, J., 2011. Gawk Scripts for Twitter Processing v1.0. Mapping Hohenstein, N.-O., Feisel, E., Hartmann, E., 2014. Human resource management
Online Publics. 〈http://mappingonlinepublics.net/http://mappingonlinepublics. issues in supply chain management research. Int. J. Phys. Distrib. Logist. Manag.
net/resources/〉 (22.06.11). 44, 434–463.
Bruns, A., Stieglitz, S., 2013. Towards more systematic Twitter analysis: metrics for Holmberg, K., Thelwall, M., 2014. Disciplinary differences in Twitter scholarly
tweeting activities. Int. J. Soc. Res. Methodol. 16, 91–108. communication. Scientometrics, 1–16. http://dx.doi.org/10.1007/s11192-014-
Burt, R., Kilduff, M., Tasselli, S., 2013. Social network analysis: foundations and 1229-3.
frontiers on advantage. Annu. Rev. Psychol. 64, 527–547. Huang, G., Zhong, R., Tsui, K., 2014. Call for Papers: Big Data for Service and
Calantone, R.J., Vickery, S.K., 2010. Introduction to the special topic forum: using Manufacturing Supply Chain Management. 〈http://www.journals.elsevier.com/
archival and secondary data sources in supply chain management research. international-journal-of-production-economics/call-for-papers/big-data-for-
J. Supply Chain Manag. 46, 3–11. service-and-manufacturing-supply-chain/〉 (accessed 15.11.13).
Carter, C.R., Ellram, L.M., Tate, W., 2007. The use of social network analysis in Hughes, A.L., Palen, L., 2009. Twitter adoption and use in mass convergence and
logistics research. J. Bus. Logist. 28, 137–168. emergency events. Int. J. Emerg. Manag. 6, 248–260.
Casemore, S., 2012. Social Media and the Coming Supply-Chain Revolution. 〈http:// Inauen, S., Schoeneborn, D., 2014. Twitter and its usage for dialogic stakeholder
www3.cfo.com/article/2012/2/supply-chain_supply-chain-innovation-social- communication by MNCs and NGOs. In: Tench, R., Sun, W., Jones, B. (Eds.),
media-casemore-ghg〉 (accessed 20.09.14). Critical Studies on Corporate Responsibility, Governance and Sustainability, vol.
Cecere, L., 2012. Big Data: Go Big or Go Home? Supply Chain Insights LLC 〈http:// 6. Emerald Group Publishing Limited, pp. 283–310.
supplychaininsights.com/wp-content/uploads/2012/07/Big_Data_Report_16_ Jansen, B., Zhang, M., 2009. Twitter power: tweets as electronic word of mouth.
JULY_2012.pdf〉 (accessed 05.11.13). J. Am. Soc. Inf. Sci. Technol. 60, 2169–2188.
Chae, B., 2012. An evolutionary framework for service innovation: insights of Java, A., Song, X., Finin, T., Tseng, B., 2007. Why we Twitter: understanding
complexity theory for service science. Int. J. Prod. Econ. 135, 813–822. microblogging usage and communities. In: Proceedings of the 9th WebKDD
Chae, B., Olson, D., 2013. Business analytics for supply chain: a dynamic-capabilities and 1st SNA-KDD 2007 Workshop on Web Mining and Social Network Analysis.
framework. Int. J. Inf. Technol. Decis. Mak. 12, 9–26. ACM, pp. 56–65.
Chae, B., Olson, D., Sheu, C., 2014. The impact of supply chain analytics on operational Kalampokis, E., Tambouris, E., Tarabanis, K., 2013. Understanding the predictive
performance: a resource-based view. Int. J. Prod. Res. 52, 4695–4710. power of social media. Internet Res. 23, 544–559.
Chau, M., Xu, J., 2012. Business intelligence in blogs: understanding consumer Kim, Y., Choi, T., Yan, T., Dooley, K., 2011. Structural investigation of supply
interactions and communities. MIS Q. 36, 1189–1216. networks: a social network analysis approach. J. Oper. Manag. 29, 194–211.
Chen, L., Olhager, J., Tang, O., 2014. Manufacturing facility location and sustain- Kumar, S., Havey, T., 2013. Before and after disaster strikes: a relief supply chain
ability: a literature review and research agenda. Int. J. Prod. Econ. 149, 154–163. decision support framework. Int. J. Prod. Econ. 145, 613–629.
 B.(K.) Chae / Int. J. Production Economics 165 (2015) 247–259 259

Lasorsa, D.L., Lewis, S.C., Holton, A.E., 2012. Normalizing Twitter. Journal. Stud. 13, Suh, B., Hong, L., Pirolli, P., Chi, E.H., 2010. Want to be retweeted? Large scale
19–36. analytics on factors impacting retweet in Twitter network. In: Proceedings of
Malhotra, A., Malhotra, C., See, A., 2012. How to get your messages retweeted. MIT the 2010 IEEE Second International Conference on Social Computing. IEEE
Sloan Manag. Rev. 53, 61–66. Computer Society, pp. 177–184.
Mann, J., 2014. Social Media Offers Opportunities for Supply Chain Management. Tang, O., Nurmaya Musa, S., 2011. Identifying risk issues and research advance-
〈http://smbp.uwaterloo.ca/2014/07/social-media-offers-opportunities-for-sup ments in supply chain risk management. Int. J. Prod. Econ. 133, 25–34.
ply-chain-management/〉 (accessed 20.09.14). Teece, D., 2007. Explicating dynamic capabilities: the nature and microfoundations
Manning, C., Raghavan, P., Schutze, H., 2008. Introduction to Information Retrieval. of (sustainable) enterprise performance. Strateg. Manag. J. 28, 1319–1350.
Cambridge University Press, Cambridge, England. Telegraph, T., 2012. Average Twitter User is an an American Woman with an iPhone
McKinsey&Company, 2011. Big Data: The Next Frontier for Innovation, Competition, and 208 Followers. 〈http://www.telegraph.co.uk/technology/news/9601327/
and Productivity. The McKinsey Global Institute, pp. 1–143.
Average-Twitter-user-is-an-an-American-woman-with-an-iPhone-and-208-fol
Miller, C., 2009. Dell Says It Has Earned $3 Million From Twitter. 〈http://bits.blogs.
lowers.html〉 (accessed 29.09.14).
nytimes.com/2009/06/12/dell-has-earned-3-million-from-twitter/?
Terry, M., 2009. Telemed. e-Health 15, 507–510. http://dx.doi.org/10.1089/tmj.2009.9955.
_php=true&_type=blogs&_r=0〉 (accessed 20.09.14).
Thelwall, M., Buckley, K., Paltoglou, G., 2011. Sentiment in Twitter events. J. Am. Soc.
Nason, G.J., O’Kelly, F., Kelly, M.E., Phelan, N., Manecksha, R.P., Lawrentschuk, N.,
Inf. Sci. Technol. 62, 406–418.
Murphy, D.G., 2014. The emerging use of Twitter by urological journals. BJU Int.
Trkman, P., McCormack, K., de Oliveira, M., Ladeira, M., 2010. The impact of business
http://dx.doi.org/10.1111/bju.12840.
Natoli, P., 2013. The Impact of Social Media on the Supply Chain: Is There One? analytics on supply chain performance. Decis. Support Syst. 49, 318–327.
〈http://blog.jda.com/the-impact-of-social-media-on-the-supply-chain-is- Tuomikangas, N., Kaipia, R., 2014. A coordination framework for sales and operations
there-one/〉 (accessed 15.05.14). planning (S&OP): synthesis from the literature. Int. J. Prod. Econ. 154, 243–262.
Naveed, N., Gottron, T., Kunegis, M., Alhadi, A.C., 2011. Bad news travel fast: a Twitter, 2013. Twitter Developer Documentation. 〈https://dev.twitter.com/docs〉.
content-based analysis of interestingness on Twitter. In: Proceedings of the 3rd Vallet-Bellmunt, T., Martínez-Fernández, M.T., Capó-Vicedo, J., 2011. Supply chain
International Web Science Conference. ACM, Koblenz, Germany, pp. 1–7. management: a multidisciplinary content analysis of vertical relations between
O’Leary, D., 2011. The use of social media in the supply chain: survey and companies, 1997–2006. Ind. Mark. Manag. 40, 1347–1367.
extensions. Intell. Syst. Account. Finance Manag. 18, 121–144. Vis, F., 2013. Twitter as a reporting tool for breaking news journalists tweeting the
Ostrow, A., 2009. Japan Earthquake Shakes Twitter Users … And Beyonce. 〈http:// 2011 UK riots. Digit. Journal. 1, 27–47.
mashable.com/2009/08/12/japan-earthquake/〉 (accessed 20.09.14). Waller, M., Fawcett, S., 2013. Data science, predictive analytics, and big data: a
Pang, B., Lee, L., 2008. Opinion mining and sentiment analysis. Found. Trends Inf. revolution that will transform supply chain design and management. J. Bus.
Retr. 2, 1–135. Logist. 34, 77–84.
Park, H., Rogers, S., Stemmie, J., 2013. Analyzing health organizations' use of Twitter Wasserman, S., Faust, K., 2005. Social Network Analysis: Methods and Applications.
for promoting health literacy. J. Health Commun. 18, 410–425. Cambridge University Press, New York.
Rogers, R., 2013. Debanalizing Twitter: the transformation of an object of study. In: Watts, D., 2003. Six Degrees: The Science of A Connected Age. W.W. Norton &
Proceedings of the 5th Annual ACM Web Science Conference. ACM, Paris, Company, New York.
France, pp. 356–365. Webster, T., 2010. Twitter Usage In America: 2010: The Edison Research/Arbitron
Sakaki, T., Okazaki, M., Matsuo, Y., 2010. Earthquake shakes Twitter users: real-time Internet and Multimedia Study. Edison Research.
event detection by social sensors. In: Proceedings of the 19th International Weiss, S., Indurkhya, N., Zhang, T., Damerau, F., 2005. Text Mining: Predictive
Conference on World Wide Web. ACM, Raleigh, North Carolina, USA, pp. 851–860. Methods for Analyzing Unstructured Information. Springer.
Seuring, S., Gold, S., 2012. Conducting content-analysis based literature reviews in Williams, S., Terras, M., Warwick, C., 2013. What do people study when they study
supply chain management. Supply Chain Manag.: Int. J. 17, 544–555. Twitter? Classifying Twitter related academic papers. J. Doc. 69, 384–410.
Seuring, S., Müller, M., 2008. From a literature review to a conceptual framework Wyatt, N., 2013. Best in Class Crisis Management with Social Media. 〈http://www.
for sustainable supply chain management. J. Clean. Prod. 16, 1699–1710. sparkcentral.com/best-class-crisis-management-social-media/〉 (accessed 12.11.13).
Smart, A., Apte, U., Davis, M., Maull, R., 2014. Information-Centric Operations and Yuan, T., Achananuparp, P., Lubis, I.N., Lo, D., Ee-Peng, L., 2012. What does software
the Digital Economy. 〈http://www.emeraldinsight.com/products/journals/call_
engineering community microblog about? In: Proceedings of the 9th IEEE
for_papers.htm?id=4757〉 (accessed 01.11.13).
Working Conference on Mining Software Repositories (MSR). pp. 247–250.
Stieglitz, S., Dang-Xuan, L., 2013. Emotions and information diffusion in social
Zappavigna, M., 2011. Ambient affiliation: a linguistic perspective on Twitter. New
media – sentiment of microblogs and sharing behavior. J. Manag. Inf. Syst. 29,
Media Soc. 13, 788–806.
217–248.