Knowledge graphs

The challenge of recognizing named entities in a given text has been a very dynamic field in recent years. This is due to the advances in neural network architectures, increase of computing power and the availability of diverse labeled datasets, which deliver pre-trained, highly accurate models. These tasks are generally focused on tagging common entities, but domain-specific use-cases require tagging custom entities which are not part of the pre-trained models. This can be solved by either fine-tuning the pre-trained models, or by training custom models. The main challenge lies in obtaining reliable labeled training and test datasets, and manual labeling would be a highly tedious task.
In this paper we present PharmKE, a text analysis platform focused on the pharmaceutical domain, which applies deep learning through several stages for thorough semantic analysis of pharmaceutical articles. It performs text classification using state-of-the-art transfer learning models, and thoroughly integrates the results obtained through a proposed methodology. The methodology is used to create accurately labeled training and test datasets, which are then used to train models for custom entity labeling tasks, centered on the pharmaceutical domain. The obtained results are compared to the fine-tuned BERT and BioBERT models trained on the same dataset. Additionally, the PharmKE platform integrates the results obtained from named entity recognition tasks to resolve co-references of entities and analyze the semantic relations in every sentence, thus setting up a baseline for additional text analysis tasks, such as question answering and fact extraction. The recognized entities are also used to expand the knowledge graph generated by DBpedia Spotlight for a given pharmaceutical text.

Knowledge graphs such as Yago and Freebase have become a powerful asset for enhancing search, and are being intensively used in both academia and industry. Many existing knowledge graphs are either available as Linked Open Data, or they can be exported as RDF datasets enhanced with background knowledge in the form of an OWL 2 ontology. Faceted search is the de facto approach for exploratory search in many online applications, and has been recently proposed as a suitable paradigm for querying RDF repositories. In this paper, we provide rigorous theoretical underpinnings for faceted search in the context of RDF-based knowledge graphs enhanced with OWL 2 ontologies. We identify well-defined fragments of SPARQL that can be naturally captured using faceted search as a query paradigm, and establish the computational complexity of answering such queries. We also study the problem of updating faceted interfaces, which is critical for guiding users in the formulation of meaningful queries during exploratory search. We have implemented our approach in a fully-fledged faceted search system, SemFacet, which we have evaluated over the Yago knowledge graph.

Graphs have become the best way we know of representing knowledge. The computing community has investigated and developed the support for managing graphs by means of digital technology. Graph databases and knowledge graphs surface as the most successful solutions to this program. The goal of this document is to provide a conceptual map of the data management tasks underlying these developments, paying particular attention to data models and query languages for graphs.

While the recent advances in Data Science and Machine Learning attract lots of attention in Cyber Security because of their promise for effective security analytics, Vulnerability Analysis, Risk Assessment and Security Policy Validation remain slightly aside. This is mainly due to the relatively slow progress in the theoretical formulation and the technological foundation of the cyber security concepts such as logical vulnerability, threats and risks. In this article we are proposing a framework for logical analysis, threat intelligence and validation of security policies in cyber systems. It is based on multi-level model, consisting of ontology of situations and actions under security threats, security policies governing the security-related activities, and graph of the transactions. The framework is validated using a set of scenarios describing the most common security threats in digital banking and a prototype of an event-driven engine for navigation through the intelligence grap...

WikiNizer™ Research (WikiNizeR) is a visual Wiki-like knowledge orgaNizer which constructs Personal Knowledge Graphs. By enabling us to visualize "meta" levels of reflection, WikiNizeR facilitates our sense making and problem solving. Using Freebase as a linked data source it harvests and contextualizes semantically structured information, empowering us to curate a private Knowledge Graph of Things, which a view also to enhancing collaborative knowledge work. Its users integrate open web data within the content driven graph architectures of their personal learning environments. It enables us to elaborate upon the emergent dynamic graphs which articulate the typed connections that exist between 'Things' of interest, facilitating the emergence of novel concepts in the associated complexes of content which are organized into node based structures. It supports a connectivist learning model; and as a "self-curating" semantic knowledge management tool it can be used in problem and project based learning to explore resources and conceptual relationships, helping us to define learning paths and workflows, or design meta-level didactic object structures and activities. It is a tool which helps us track meaning construction in a situated, intent dependent and dynamic manner. It is a holistic solution which integrates web research, linked data, annotation, note-taking and knowledge organization into a Lifelong Personal Digital Archive of "born reproducible", ab initio re-purposable, and re-enactable, Research Objects. Track: Open Track

Art looting networks operate on many levels, many of them hidden, over long periods of time. Documenting and visualizing the connections - and linking them across disparate sources of information - are tasks that graphs are particularly well suited for. This paper describes efforts to track Nazi-art looting networks using the native graph functions of Wikidata enhanced by federated queries.

Twitter users get the latest tweets of their followees on their timeline. However, they are often overwhelmed by the large number of tweets, which makes it difficult for them to find interesting information among them. In this work, we present an efficient semantic recommendation method that helps users filter the Twitter stream for interesting content. The foundation of this method is a knowledge graph (KG) that can represent all user topics of interest as a variety of concepts, objects, events, persons, entities, locations and the relations between them. Our method uses the KG and graph theory algorithms not yet applied in social network analysis in order to construct user interest profiles by retrieving semantic information from tweets. Next, it produces ranked tweet recommendations. In addition, we use the KG to calculate interest similarity between users, and we present a followee recommender based on the same underlying principles. An important advantage of our method is that it reduces the effects of problems such as over-recommendation and over-specialization. As another advantage, our method is not impaired by the limitations posed by Twitter on the availability of the user graph data. We implemented from scratch the best-known state-of-the-art approaches in order to compare with them and assess our method. Moreover, we evaluate the efficiency and runtime scalability of our method.

Student behavior analysis plays an increasingly important role in education data mining research, but it lacks systematic analysis and summary. Based on reading a large amount of literature, this paper has carried out the overall framework, methods and applications of its research. Comprehensive combing and elaboration. Firstly, statistical analysis and knowledge map analysis of the relevant literature on student behavior analysis in the CNKI database are carried out, and then the research trends and research hot spots are obtained. Then, from the different perspectives of the overall process and technical support of student behavior analysis, the overall framework of the research is constructed, and the student behavior evaluation indicators, student portraits and used tools and methods are highlighted. Finally, it summarizes the principal applications of student behavior analysis and points out the future research direction.

Twitter is a rapidly growing microblogging platform that allows its users to send and read short messages, called tweets. Because of the fact that a user's timeline consists of the latest tweets of their followees (users that they are following), followee recommendation is a problem of significant importance. In this work, we propose a followee recommendation approach, which takes advantage of the increasing amount of available social data and specifically the semantic relatedness of topics that interest users. In order to accomplish this, we use a Topic Graph, containing a wide variety of topics that will be used for the recommendation process. Today knowledge graphs provide a solid basis for us to construct a full and reliable Topic Graph. Our approach takes advantage of the semantic information retrieved from users' tweets, in order to build an interest profile for each user. Then we use graph theory algorithms in order to calculate user interest similarity using the Topic Graph.

The Semantic representation plays a vital role in
information extraction and information retrieval. Current
state-of-the-art for semantic representation are Temporal
Knowledge Graphs. Due to tremendous growth of things
like Big Data we need a way to devise how we should
represent available knowledge such that it can be more
effective. Finding out semantically connected things from
huge data lake is complex task which comes under the
umbrella of Computer Science specifically semantic web.
In this document we have presented a novel temporal
Knowledge Graph dataset. Proposed dataset encompasses
ten different types of relationship among which two
relations are temporal based. Dataset incorporate temporal
data of hundred people. For data acquisition we have used
university campus environment. By using facial
recognition technique backed with Convolutional Neural
Network (CNN) [1] we have collected data of each person.
For training our model on facial features we have collected
approx. 500 images of each person. Afterward for profiling
people we deployed surveillance cameras on different
locations of campus. Finally, acquisitioned data has been
transformed to Temporal Knowledge Graph. We will
discuss data collection process details in dedicated section.
Knowledge Graph is simply a graph representing
interconnections between entities [2]. For analysis of
dataset we have modelled some semantic queries. Some
sample queries are as under:
Where were XYZ person went all the day today?
Which routes Prof. usually takes in going to department office?
How many students were present on main gate in today at XYZ
time today?
How many people were present on XYZ seminar/ workshop?
How many people were present in the Research Lab when shortcircuit happened.
Which places Alice visited today?
Which areas of campus faced Student congestion today?
How many students attended the particular event on specific
date?
How many faculty members were present in today’s meeting?
How many ambulances/ fire fighters reached on time when
certain event occurred?
How many people joined in the opening ceremony of
department/ Lab?
How many people visited the particular department/
departments?
What are the peak time of admission in terms of students?
How many and which departments Alice visited in her last visit.
Which library/ department Bob visits most often?
How many students visited accounts office today?
How many students visited Chancellor/ Vice Chancellor office
today?
How many people were at Cafeteria at XYZ time?
How many students were present at playground today?
How many students were present at gymnasium today?
The list is not exhaustive and many questions related to
current and past events are required to be probed while
future can be predicted. Proposed dataset can be used in
different domains like citizen profiling, workplace
profiling, parental monitoring, student profiling,
autonomous vehicle profiling etc.

Graphs have become the best way we know of representing knowledge. The computing community has investigated and developed the support for managing graphs by means of digital technology. Graph databases and knowledge graphs surface as the most successful solutions to this program. The goal of this document is to provide a conceptual map of the data management tasks underlying these developments, paying particular attention to data models and query languages for graphs.

The amount of video content available on the Web is constantly growing, especially due to the increasing popularity of Video on Demand (VoD) platforms such as Netflix, Hulu and Youtube. This has made it harder for viewers to discover the right visual content for them. Recommender systems are being offered by VoD services in order to automatically suggest potentially interesting videos to users. However, recommendations are typically based on: (i) limited video metadata fields, such as genre, title and actors; (ii) the content that other users liked; (iii) private or isolated data repositories. In this paper we describe our approach leveraging the richness of semantic technologies to enrich movie metadata, and create meaningful semantically-enriched descriptions of movie scenes using video and audio processing techniques. This approach allows for the creation of a structured and interoperable Knowledge Graph (KG) describing movies and their content. This KG can be easily interlinked with existing Linked Data datasets available on the Web, resulting in a more comprehensive representation of the movies. Our approach enables seamless integration with existing data sources and fine-grained data analysis of movie content at Web scale.

Patterns for the internet of things (IoT) which represent proven solutions used to solve design problems in the IoT are numerous. Similar to objectoriented design patterns, these IoT patterns contain multiple mutual heterogeneous relationships. However, these pattern relationships are hidden and virtually unidentified in most documents. In this paper, we use machine learning techniques to automatically mine knowledge graphs to map these relationships between several IoT patterns. The end result is a semantic knowledge graph database which outlines patterns as vertices and their relations as edges. We have identified four main relationships between the IoT patterns-a pattern is similar to another pattern if it addresses the same use case problem, a large-scale pattern uses a small-scale pattern in a lower level layer, a large pattern is composed of multiple smaller scale patterns underneath it, and patterns complement and combine with each other to resolve a given use case problem. Our results show some promising prospects towards the use of machine learning techniques to generate an automated repository to organise the IoT patterns, which are usually extracted at various levels of abstraction and granularity.

Finding  similar  entities  among knowledge  graphs  is  an  essential  research  problem  for knowledge  integration  and  knowledge  graph  connection.  This  paper  aims  at  finding semantically similar entities between two knowledge graphs. This can help end users and search agents more effectively and easily access pertinent information across knowledge graphs. Given a  query  entity  in  one  knowledge  graph,  the  proposed  approach  tries  to  find  the  most  similar entity  in  another knowledge  graph.  The  main idea  is to leverage graph embedding, clustering, regression and sentence embedding. In this approach, RDF2Vec has been employed to generate vector  representations of  all  entities  of  the  second  knowledge  graph  and  then the  vectors  have been clustered based on cosine similarity using K medoids algorithm. Then, an artificial neural network with multilayer perception topology has been used as a regression model to predict the corresponding vector in the second knowledge graph for a given vector from the first knowledge graph. After determining the cluster of the predicated vector, the entities of the detected cluster are ranked through sentence-BERT method and finally the entity with the highest rank is chosen as the most similar one. To evaluate the proposed approach, experiments have been conducted on  real-world  knowledge  graphs.  The  experimental  results demonstrate  the  effectiveness  of  the proposed approach.

The problem of querying RDF data is a central issue for the development of the Semantic Web. The query language SPARQL has become the standard language for querying RDF since its W3C standardization in 2008. However, the 2008 version of this language missed some important functionalities: reasoning capabilities to deal with RDFS and OWL vocabularies, navigational capabilities to exploit the graph structure of RDF data, and a general form of recursion much needed to express some natural queries. To overcome these limitations, a new version of SPARQL, called SPARQL 1.1, was released in 2013, which includes entailment regimes for RDFS and OWL vocabularies, and a mechanism to express navigation patterns through regular expressions. Unfortunately, there are a number of useful navigation patterns that cannot be expressed in SPARQL 1.1, and the language lacks a general mechanism to express recursive queries. To the best of our knowledge, no efficient RDF query language that combines the above functionalities is known. It is the aim of this work to fill this gap. To this end, we focus on a core fragment of the OWL 2 QL profile of OWL 2 and show that every SPARQL query enriched with the above features can be naturally translated into a query expressed in a language that is based on an extension of Datalog, which allows for value invention and stratified negation. However, the query evaluation problem for this language is highly intractable, which is not surprising since it is expressive enough to encode some inherently hard queries. We identify a natural fragment of it, and we show it to be tractable and powerful enough to define SPARQL queries enhanced with the desired functionalities.

Collective Intelligence is an immense research area that has wide application to cross-disciplines, like social, legal, and computation. Research trends in Surveillance find its place in the work of this area generating curated data set helpful in answering complex queries. Sousveillance is a recent term coined by researchers and had been discussed in different literatures. However our findings suggest that integration of Surviellance through Sousviellance data set has not been given much importance in collective fashion. In this literature we introduced an effective model of collective intelligence by integrating surveillance through sousviellance in a campus environment. For testbed networking devices are used to generate sousvillance data to provide validation, and cleaning to enable reliability and trust in the target object.

Ιn this talk, I present Lekythos. Named after the namesake ancient Greek clay vessels for oil, Lekythos aims to: first, represent knowledge in the domain of ancient Greek pottery by building a model of ancient Greek pottery concepts. Second, publish these concepts in the form of a multilingual knowledge graph (ontology). Third, publish an openly accessible electronic dictionary of the English, French, and Greek terms denoting these concepts.
I will discuss in length how Lekythos was built with Tedi, a software platform for building multilingual knowledge graphs. My claim is that, in contrast to Protégé, the standard editor for knowledge modeling, Tedi is more user-friendly. When building a graph in Tedi, the digital humanist can be guided by the terms used in one or more natural languages. Importantly, the concepts (classes) in Tedi are defined without the modeler having to come to grips with the steep learning curve of formalisms such as property restrictions. Using Tedi neither assumes nor requires any background in Description Logics, while it exports RDF/OWL files that can be opened in Protégé for visualisations, Sparql queries, metrics etc. As multilingual knowledge graphs are powerful software artefacts for organizing knowledge, linking data, annotating texts and images, their use can help increase dataset interoperability and knowledge discovery in Digital Humanities. To satisfy these needs, I claim, user-friendliness and a low threshold are crucial.

advancements in research tools and databases have accelerated the scientific research life cycle. However, the chronological gap between published research, research in progress and emerging research topics is shrinking, thus putting pressure on researchers to find novel research ideas. The Literature Review (LR) process is a fundamental process that can identify gaps in the research literature and stimulate new research ideas. While many researchers adopt different methodologies conducting LR, there is no methodology that can comprehensively unveil innovative research ideas. This research aims to develop a search by concepts framework. The framework involves the use of Natural Language Processing (NLP), Knowledge Graphs (KGs), and Question Answering systems (QA) to ease finding relevant concepts related to a certain scientific topic along with associated files and citations that would in return maximize the efficiency of the scientific research. The framework also allows researchers to visualize the connection between different concepts similar to the cognitive imaging of the human mind.

Within the last decade, there has been considerable feats in technology, and this has brought about low-cost innovations that were in the past only theoretical. Today, we see miniaturized super-fast computers, big data being transmitted over long distance via wireless links in split of seconds, things talking to things, amongst others. All these have paved way for the envisioned self-driving cars to emerge from the drawing board and being seen on our roads. But this innovation, which would have been one of the best approaches to mitigate road accidents and bring safety to our roads have been marred by so many roadblocks. This research looked at how various persons in Sydney, Australia react to the use of self-driving cars and it was deduced that one of the major roadblocks to adoption of self-driving cars in Sydney, Australia is human psychology. And furthermore, proposed how this problem could be mitigated.