International Journal of Recent Technology and Engineering (IJRTE)
ISSN: 2277-3878, Volume-8 Issue-5, January 2020
An Efficient Mining for Recommendation
System for Academics
Nikhat Akhtar, Devendera Agarwal
o
Abstract: At present time huge numbers of research articles
are available on World Wide Web in any domain. The research
scholar explores a research papers to get the appropriate
information and it takes time and effort of the researcher. In this
scenario, there is the need for a researcher to search a research
based on its research article. In the present paper a method of
Knowledge ablation from a collection of research articles, is
presented to evolve a system research paper recommendation
system (RPRS), which would generate the recommendations for
research article based on researcher choice. The RPRS
accumulate the knowledge ablated from the pertinent research
articles in the form of semantic tree. It accumulates all the literal
sub parts with their reckoning in nodes. These parts are arranged
based on their types in such a way that the leaf nodes stores the
words with its prospect, the higher layer gives details about dictum
with its reckoning, next to it an abstract. A Bayesian network is
applied to construct a verisimilitude model which would quotation
the pertinent tidings from the knowledge tree to construct the
recommendation and word would be scored through TF-IDF
value.
Index Terms: Bayesian Network, Text Clustering, Knowledge
Extraction, Text Classification, Recommender System
I. INTRODUCTION
Recommender systems are [1] one of the most [2]
common and convenient conceivable [3] applications of Big
data [4]. Presently, intense growth in the amount of
information is available it is time consuming to explore the
information of great value from the collective information
[5]. This occurrence is called information overload. The
information overload actually implies the availability of too
much data or information that is beyond the manageable
limits of the user and underlay big nodes in all sorts of
decision makings. This onerous occurs mainly, when the
system is unable to manage and process this huge amount of
information in an orderly manner. Because of this, in many
e-commerce applications, commonly the user has a plenty of
options, but with a very limited time to enucleate them all. A
recommendation system, the most magnificent mechanism in
this direction, effort to tackle the information overload issue,
and its subclass of information filtering system [4] that try to
quest to prediction ‘assessment’ or ‘precedence’ [1] that a
user would give to an item for instance research paper, books,
music, movies or social element they had not yet considered.
On the basics of previous effectuation of recommendation
systems, there are two main elemental [5] prescripts to this
Revised Manuscript Received on January 15, 2020
Nikhat Akhtar, Research Scholar Ph.D (Computer Science &
Engineering), Babu Banarasi Das University, Lucknow, India
Prof. (Dr.) Devendera Agarwal, Professor, Babu Banarasi Das
University, Lucknow, India
Retrieval Number: E5924018520/2020©BEIESP
DOI:10.35940/ijrte.E5924.018520
hypostasis prescript and immaterial prescript. The hypostasis
methods are analytical in character and immensely
extensively used. These perspectives rank the phrase based
on some important extent to devise the result [6]. In this
paper, we discuss the generic hypostasis prescript to
engender recommendations for conglomeration of associated
corpus [7].
It is demonstrated tentatively that time decay in perusal is
highly correspond with the organization of abstract and
phrase [4]. This RPRS decomposes the process into two
distinct sub processes. In the first process extraction of
semantic knowledge of the corpus is done. In the second
process, facts are going to be applied for the recommendation
via Bayesian network [6] of word and abstract. The
introduced RPRS are envisaged in elaboration in the
subsequent section.
II. BACKGROUND
Recommender systems are increasingly used on the Web to
assist users to discover material relevant to their interests.
The first research article on recommender system was
published in 1998. After that an impressive number of
research articles had been published. The many circumstances
have been explained to progress the credibility of
recommender system. The Recommendation systems endow
a promising method to ranking research article according to a
user’s choice [8]. The most vital feature of a recommender
system is its proficiency to “supposition” a user’s liking and
interests by inspecting the demeanor of this user and & or the
demeanor of other users to originate personalized
recommendations [9]. In the [10], authors have talk about
the offline and online evaluation of research paper
recommender model and conclude that offline evaluation in
this domain does not provide potential outcome. Yang et al.
[11] presented a recommendation system for sciential papers
that used a class oriented collaborative filtering method.
However, their system overcomes the cold-start matter by
utilizing implicit behaviors extracted from a subscriber
access log web usage data are noisy and not genuine
generally as pointed out in. Again research paper
recommender model is introduced using a Dynamic
Normalized Tree of Concepts (DNTC) model and a
complicated ontology [12]. The system is evaluated offline
using ACM digital library papers and the outcome show
that this model performs preferable than the vector of space
framework. Ekstrand et al. [13] introduced the two phases by
running a content-based filtering and a collaborative based
filtering recommender in parallel and admixture the resulting
classed lists.
1619
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
An Efficient Mining for Recommendation System for Academics
The first items on the modulate recommendation list are
those items which appeared on together lists, ordered by the
sum of their class. An astonishingly collaborative filtering
outperforms all hybrid algorithms in their assessment.
MKu¨cu¨ktunc et al. [14] developed a personalized paper
recommendation service, called the Advisor3, which
permission a subscriber to delineate her search toward
neoteric progress research papers utilize a flank conscious
unsystematic itinerancy with restart algorithm. The
recommended articles regress by the Advisor is
miscellaneous by parameterized ease up on spatial utmost.
Wang et al. [15] proposes to include textual information to
construct a topic model of the research papers and adds an
additional unnoticeable variable to differentiate between the
focus of a research paper and the domain it just talks about.
Sholom M. al. [16] introduced the algorithm in retrieving the
most pertinent documents, meanwhile precision shows the
capacity of the framework in excluding not interesting
documents. Instantly the model’s predictive performance is
adjusted, the final phase consists of presenting new and
unseen data and gets the concluding result of the
classification model. Caragea et al. [17] introduced the matter
of quotation recommendation utilize eccentric significance
putrefaction of the contiguity matrix related with the
quotation graph to build an abstruse semantic stead a
deficient dimensional stead where consonant in research
papers can be more dispassionately identified. The authors
have proposed [18] keywords based retrieval procedure in
for giving an overview and a multiple arrangement of
papers as a piece of the preparatory reading index.
K. El-Ariniet al. [19] proposed by returns a set of reasonable
research articles by optimizing an objective function based
on a fine grained notion of impact between research papers
and it says that, research article recommendation, elucidate a
query as a small cluster of familiar with to be appropriate
research papers are superior than a string of keywords.
Authors have discussed [20] the performance of stereotype
and most renowned recommendations in the area of
scholarly recommender frameworks. W. Huang et al. [21]
proposed the vocabulary used in the citation reference and in
the content of the research article is absolutely dissimilar. To
address this drawback, and presented to use translation
model, which can viaduct the gap between two different
mother tongue. The assortment performance is the simple
performance of implementing assortment models [22]. X. Liu
et al. [23] proposed citation recommendation from the
miscellaneous network mining perspective has attracted more
scrutiny. Moreover, research papers, metadata such as
authors or keywords are also to be of opinion as entities in the
graph schema. Tarus et al. [24] proposed to literature review
is presented on ontology based recommender frameworks in
the area of e-learning.
III. PROPOSED METHODOLOGY
A. Overview
At the beginning, a corpora has generated by amassing
several research periodicals concerned to dissimilar domains.
A conceptual perception is fabricated by made up of
weighted textual content of all the research materials
discovered pertinent to the given scholar’s sample abstract.
The pertinent papers in other words, conglomerations of
Retrieval Number: E5924018520/2020©BEIESP
DOI:10.35940/ijrte.E5924.018520
homogeneous research periodicals are chosen based on
prioritize their abstract using Naive Bayes classifiers.
Subsequently, the abridged is going to be procreated by
enforce a suggested hypostasis procedure on the
conglomerated datum derived from the scholar’s abstract.
Generate the researcher’s thought by applying Naive Bayes
NB10 method, entire set of research periodicals are prioritize
and the uppermost would be chosen in the recommendation
process. The research paper’s abstract materials are used in
the procedure of prioritize [25]. The scoring is calculated by a
Bayesian network technique. Thereafter, rank of abstract is
contemplating to construct the recommendation.
To keeping in mind an abstract has characteristic of
representing the entire content of a research paper, a
recommendation list is generated from the selected research
papers abstract [26]. Lastly a recommendation list retrieves
and integrate abstract alongside entire the research papers.
The sectionalisation of research periodicals is needed to
divide into feasible fragment for creates the conceptual
perception. The greater number of the research periodicals
sequence in the segments of research periodicals for instance
the name of a research paper abstract, introduction, related
work, proposed method, results and last conclusions. In
universally a lot of research papers are not accessible freely
for example (IEEE, Science Direct, and ACM) in full text
only research paper abstract is accessible.
Pseudo code
Procedure RPRS
sSet
k
While sSet
s
k
pop
k
t, S
S
NB10
do
sSet
ABS s
end while
K score
RC
BN
k
RankAbs K score
l
While RCSet
pop max
RCSet
do
T title (RC)
T
l
l
end while
return l
end procedure
In this scenario, we will use only abstract in our proposed
idea. That will utilize in the process of construction of
conceptual perception that will play the role of essential
constituents for the recommendation formation. Through
hypostasis manner [26] the significant portion of our work is
that the presentation of matter of fact, research paper literal
elements are being ranked. The result of previous stage, at the
moment we need to be stored in the conceptual perception, so
that appropriate prioritization of all research paper wording
identical to dictum extant at differing schema of research
paper performed. The objective of the present paper is to
construct a list that will provide the title of the research paper.
1620
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
International Journal of Recent Technology and Engineering (IJRTE)
ISSN: 2277-3878, Volume-8 Issue-5, January 2020
B. The Knowledge Tree formation
Certain features are desirable for the perception delineation
[27] in terms of apparent, emphatic, better expressiveness,
dexterous, influential, and independent of the circumstances
[28]. The tree much robust knowledge base [29] that can
accumulate the regain intimation from the collection of
research paper. It’s completely arduous to split the document
if the limitations of segments are not explicit. In this context,
stream of words are not identical via whole research paper
script. For the suitable intellect eduction, research paper
script becomes split suitably at the point where word
transformation is high [30]. As the research documents we
are previously split into many segments, and split according
to the order of headings.
The knowledge is extracted from the abstracts of all the
research paper and become visible for effortlessly
conceivable and able to be used in our tree structure. After
that arranged in order of delineation intellect is utilized to
conglomeration the similar research paper documents, on the
basis of category of research paper documents be related to
shown in figure 1. This amalgamates intellect framework
applying for the creation of an accurate and pervasive
knowledge. The objective of this circumstance arranged
intellect delineation is the [31] concernment of research
paper textual units is comfortably roughly calculated in
bottom-top manner. Afterward the generation of a tree every
research paper script is split into constituent ahead into
dictum. This procedure requires recurring until vocabulary’s
level of Eduction accomplished triumphantly. In this
framework word would be scored firstly through term
frequency inverse document frequency (tf-idf) value.
graph (DAG). In this first component Bayesian network B
would be defined to represent a 2 tuple < D,λ > where D is the
directed acyclic graph whose nodes are indiscriminate
variable N1, N2, N3, . . . , Nn and where connections indicate
provisory reliance among [32] arbitrary variables. The
secondly components condition, likelihood table for every
variable. In this component of the pair λ indicates the
conglomeration of that reliance likelihood amidst the
vertices. The conglomeration includes measurable factor λ Ni
|σi= LB (Ni |σi) showing likelihood of Ni conditioned on σi.
Figure 2: The Bayesian Network Construct among
Word Vertices & Abstract Vertices
A Bayesian network B is procreated among abstract &
word provided in figure 2. Here upon these network
procreated via linking the abstract vertices to word vertices, if
the particular word be suited with that abstract. Again, Tf-idf
of word is applied for the calculation of the edges’ weight to
spruce the next factor λ of the Bayesian network. The
likelihood of the abstract is count up in the given expression.
k
LB ( L j ) L j i
i 1
In training procedure it reads the text from research paper
sentence by sentence, discover the count of feature words and
differentiate with the abstract stored. Again, it computes the
frequency of every feature word. In the check-up procedures
it discovers the counts of feature words to categories
according to the research paper.
IV. RECOMMENDATION RESULTS &
EVALUATION
Figure 1: Knowledge Tree framework
C. Bayesian Network Based Scoring
The research paper abstract choices are done with a Bayesian
network of this research paper abstract. Bayesian networks
are probabilistic models that clearly take over the familiar
conditional dependence with directed edges in a particular
graph ideality and this ideality can demonstrate correlation
between variables and used to calculate likelihood. These
network models take over both conditionally dependent and
conditionally independent relationships between random
variables. Consequently, the connections among nodes point
out the likelihood reliance among them. Basically, these
networks have two components, firstly the directed acyclic
Retrieval Number: E5924018520/2020©BEIESP
DOI:10.35940/ijrte.E5924.018520
The presented RPRS method is computed by a repository that
has been made by selection some research paper. For
instance research papers domain like Artificial Intelligence,
Network Security, Data Mining, Big Data, and Document
Recommendation. Every domain is constructed by the
collection of same kind of research papers. Based on the
likeness with given research articles, appropriate research
articles would be chosen from the knowledge tree and used
for recommendation purpose. From the classified cluster of
research paper that contain research paper from various field
like Artificial Intelligence, Network Security, Data Mining,
Big Data [4][33], and Document Recommendation
recommender system the domain of paper is identify by using
Bayesian classification. After this identification process a set
of research papers abstract is used for the ranking process by
using Naïve based classification NB10 approach.
1621
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
An Efficient Mining for Recommendation System for Academics
The steps of this ranking process are given as follows. We
have considered 5 abstract [34][35][36][37][38]. For the sake
of simplicity 4 abstract are used in training the system and
one abstract is used for testing the system.
Step 1 Data ablation technique is applied on all 4 training
research paper abstract to consider only conceptual word
means delimiters and common words like (the ,is, there, etc)
are removed.
Step 2 The tables for words are create that include word count
of each unique word.
Table- I: The Bag of the Words Representation Paper 1
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Words
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
1
1
2
8
2
3
2
2
2
1
1
1
1
2
1
1
2
1
1
1
1
2
2
2
1
1
2
1
1
1
1
Retrieval Number: E5924018520/2020©BEIESP
DOI:10.35940/ijrte.E5924.018520
Total
Counts = 35
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Words
Study
Dictionary based
Method
Extract
Expressive
Concepts
Document
Concept mining
English
Turkish
Agglutinative
Language
Immature
Dictionary
WordNet
Lexical
Database
Grouping
Word
Synsets
Synonyms
Hypernyms
Hyponyms
Relationships
Meaning
Text
Success
High
Determining
Collected
Corpora
Counts
3
1
2
2
1
2
2
1
1
1
1
1
1
2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
Total
Counts = 38
Table- IV: The Bag of the Words Representation
Paper 4
No.
Total
Counts = 61
Words
No.
Vocabulary = 31
2
1
1
1
1
1
1
1
1
Text Mining
Techniques
Statistical analysis
Terms
Enhance
Document
Sentence
Mining
Corpus based
Concept based
Clustering
Text
Model
Important
Concept based mining
Analysis
Quality
Corpus
Approach
Corpus level
Sentence based
Concept analysis
1
1
1
Table- III: The Bag of the Words Representation Paper 3
Table- II: The Bag of the Words Representation Paper
2
No.
Similarity
Measure
Document based
Vocabulary = 25
Counts
Text Mining
Techniques
Statistical analysis
Word, term, phrase
Frequency
Document
Same
Meaning
Contributes
Appropriate
Semantics
Text
More
Importance
Concept -based mining
Analyses
Terms based
Sentence level
Document level
Corpus level
Sentence based
Concept analysis
Term frequency
Document term frequency
Document based
Corpus based
Document frequency
Concept based
Similarly
Corpus
Concept-Based Analysis
Algorithm
Clusters
Web
Efficient
Quality
Model
Significantly
Traditional
Single -term-base
Approaches
Vocabulary = 40
23
24
25
Counts
1
1
2
3
1
3
2
2
1
1
1
1
2
1
1
2
1
1
1
1
1
2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Words
Quick
Growth
Information
Extract
Discover
Knowledge
Data sources
World Wide Web
Methods
Text mining
Statistical analysis
Phrase or word
Document
Bags of words
Importance
Meaning
Content
Term
Frequency
Significance
Within
Same
Pays
Sentence
Concept -based
Analyses
Planned
Model
Clustering
K-means
Similarity
Vocabulary = 31
1622
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
Counts
1
1
1
2
1
1
1
1
2
1
2
1
5
1
1
2
1
5
2
1
1
1
1
1
3
2
1
2
1
1
1
Total
Counts = 48
International Journal of Recent Technology and Engineering (IJRTE)
ISSN: 2277-3878, Volume-8 Issue-5, January 2020
Step 3 Simultaneously a common list that contain all unique
words from all the research papers abstracts is also generated
that is called vocabulary of the corpora.
78
79
80
81
82
83
Table- V: The Common List Vocabulary of the
Corpora
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
Words
Text Mining
Techniques
Statistical analysis
Word, term, phrase
Frequency
Document
Same
Meaning
Contributes
Appropriate
Semantics
Text
More
Importance
Concept -based mining
Analyses
Terms based
Sentence level
Document level
Corpus level
Sentence based
Concept analysis
Term frequency
Document term frequency
Document based
Corpus based
Document frequency
Concept based
Similarly
Corpus
Concept-Based Analysis Algorithm
Clusters
Web
Efficient
Quality
Model
Significantly
Traditional
Single -term-base
Approaches
Mining
Sentence
Measure
Enhance
Study
Dictionary based
Method
Extract
Expressive
Concepts
Concept mining
English
Turkish
Agglutinative
Language
Immature
Dictionary
WordNet
Lexical
Database
Grouping
Synsets
Synonyms
Hypernyms
Hyponyms
Relationships
Success
High
Determining
Collected
Quick
Growth
Information
Discover
Knowledge
Data sources
World Wide Web
Retrieval Number: E5924018520/2020©BEIESP
DOI:10.35940/ijrte.E5924.018520
Bags of words
Content
Within
Pays
Planned
K -means
Vocabulary = 83
Step 4 The probabilities of each research paper document is
computed.
p of Set s
Entire number of training phrase belonging to Set s
Entire number of phrase in the training set
Table- VI: The Training abstracts probability p of set “s”
Class c
D1
D2
D3
D4
Training abstracts probability p of set s
10
5
5
7
10/(27)= 0.37037
5/(27)= 0.18518
5/(27)= 0.18518
7/(27)= 0.25925
Step 5 After that the verifying abstract is considered. For the
sake of ingenuity we are using only one longest sentence.
Step 6 After text ablation techniques only certain words will
identified these words are used for the computation task.
Step 7 In subsequent computation we find the ranking of all
abstract in these ranking the highest certain papers are
recommended to the user.
1623
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
An Efficient Mining for Recommendation System for Academics
Table- VII: The Enumeration of Assay Term in “k” in Set “s”
Enumeration of Assay Term in “k” in Set “s”
Class c
Concept Based
2
2
1
2
D1
D2
D3
D4
Approach
2
2
1
1
Search
1
1
1
1
Information
1
1
1
2
Term
9
4
2
1
Meaning
3
1
1
2
Present
1
1
1
1
Keyword
1
1
1
1
Text
2
2
1
1
Table- VIII: The Probability of Assay Term in “k” in Set “s”
Probability of Assay Term in “k” in Set “s”
Class c
Concept
Based
0.01379
0.01680
0.00819
0.01515
D1
D2
D3
D4
Approach
Search
Information
Term
Meaning
Present
Keyword
Text
0.01379
0.01680
0.00819
0.00757
0.00689
0.00840
0.00819
0.00757
0.00689
0.00840
0.00819
0.01515
0.06200
0.03361
0.01639
0.00757
0.02068
0.00840
0.00819
0.01515
0.00689
0.00840
0.00819
0.00757
0.00689
0.00840
0.00819
0.00757
0.01379
0.01680
0.00819
0.00757
Table- IX: The Multiplication M of Assay Term in “k” in Respective Set
Multiplication [M of Assay Term in “k” in Set Propitious s] = M(Concept Based) * M (Approach) * M (Search)
* M (Information) * M (Term)* M (Meaning) * M (Present) * M (Keyword) * M (Text)
Class c
Concept
Based
0.01379
0.01680
0.00819
0.01515
D1
D2
D3
D4
Multiplication
Approach
Search
Information
Term
Meaning
Present
Keyword
Text
0.01379
0.01680
0.00819
0.00757
0.00689
0.00840
0.00819
0.00757
0.00689
0.00840
0.00819
0.01515
0.06200
0.03361
0.01639
0.00757
0.02068
0.00840
0.00819
0.01515
0.00689
0.00840
0.00819
0.00757
0.00689
0.00840
0.00819
0.00757
0.01379
0.01680
0.00819
0.00757
7.57725512E-18
6.66488696E-18
3.31779973E-19
6.54354624E-19
Table- X: The r related to Set “s”
M [ r related to Set “s”] = Multiplication [M of Assay Term in “k” in Set s] * M of Set s
Class c
Concept
Based
0.01379
0.01680
0.00819
0.01515
D1
D2
D3
D4
Approach
Search
Information
Term
Meaning
Present
Keyword
Text
Multiplication
0.01379
0.01680
0.00819
0.00757
0.00689
0.00840
0.00819
0.00757
0.00689
0.00840
0.00819
0.01515
0.06200
0.03361
0.01639
0.00757
0.02068
0.00840
0.00819
0.01515
0.00689
0.00840
0.00819
0.00757
0.00689
0.00840
0.00819
0.00757
0.01379
0.01680
0.00819
0.00757
2.80638798E-18
1.23423709E-18
6.14390154E-20
1.69641436E-19
Step 8 In our example among 4 abstract 3 are used for
recommendation process and lowest rank is left because it is
least relevant.
With the same procedure we have generate recommendation
for other subject field research papers. The performance
measures namely as precision, recall, and -score are
calculated for the introduced technique. Precision estimates
the chastity percentage and recall measures the perfection of
the summarizer and uniting of precision and recall, -score is
evaluated.
Precision
Recall
relevant research paper retrieved research paper
retrieved research paper
relevant research paper retrieved research paper
relevant research paper
F-Measure
Recall * 2*Precision
Recall Precision
For Data Mining, Precision = 34/ (34+2) = 0.9444
2. The values of recall are calculated as,
For Big Data, Recall = 36/ (36+10) =0.7826
For Artificial Intelligence, Recall =45/ (45+6) = 0.8823
For Document Recommendation, Recall = 50/ (50+15) =
0.7936
For Network security, Recall = 32/ (32+8) = 0.8000
For Data Mining, Recall = 34/ (34+8) = 0.8947
3. The values of f-measure are calculated as,
For Big Data, f-measure = 1.3742/1.6606= 0.8275
For Artificial Intelligence, f-measure = 1.6543/1.8198=
0.9090
For Document Recommendation, f-measure =
1.4427/1.7026= 0.8473
For Network security, f-measure = 1.4627/1.7142= 0.8532
For Data Mining, f-measure = 1.6899/1.8391= 0.9188
Table- XI: Knowledge Base Retrieved Research Paper
No
.
Knowledge Base Area
1
2
3
Big Data
Artificial Intelligence
Document
Recommendation
Network Security
Data Mining
4
5
Retrieved
Not
Retrieved
Not
Pertinent
36
45
50
10
6
13
5
3
5
Not
Pertinent
not
Retrieved
49
46
32
32
34
8
4
3
2
57
60
Table- XI: The Result of Precision, Recall, F- Measure
No.
1
2
3
4
5
Knowledge Base Area
Precision
Big Data
Artificial Intelligence
Document
Recommendation
Network Security
Data Mining
0.8780
0.9375
0.9090
0.7826
0.8823
0.7936
0.8275
0.9090
0.8473
0.9142
0.9444
0.8000
0.8947
0.8532
0.9188
1. The values of precision are calculated as,
For Big Data, Precision = 36/ (36+5) = 0.8780
For Artificial Intelligence, Precision = 45/ (45+3) = 0.9375
For Document Recommendation, Precision = 50/ (50+5) =
0.9090
For Network security, Precision = 32/ (32+3) = 0.9142
Retrieval Number: E5924018520/2020©BEIESP
DOI:10.35940/ijrte.E5924.018520
1624
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
Recall
F-measure
International Journal of Recent Technology and Engineering (IJRTE)
ISSN: 2277-3878, Volume-8 Issue-5, January 2020
7.
8.
9.
10.
Figure 3: The Performance Evaluation of Precision,
Recall, F- Measure Outcome
11.
The X-axis represents the knowledge base of the research
paper like Artificial Intelligence, Network Security, Data
Mining, Big Data, and Document Recommendation and the
Y-axis indicates the scale of intervals. If recall, f- measure,
precision reaches near 1 it shows that the system is more
effective.
13.
V. CONCLUSION
15.
In this paper we have introduced the Bayesian classification
method for recommending research paper. This method is
favorable for our purpose. As it performed a better
classification method so we are easily able to identify the
class of the given paper, then by using a naïve base NB10
approach we have computed the rank of selected categories’
papers and highest rank papers are recommended to the user.
The paper uses only abstract that are most of the time publicly
available as compare to other baseline method that uses
whole text that are not always publicly available. Our
proposed method has presented the expressive reformation
better than baseline method of evaluate the executions and
potency to retrieve legitimate result with advantageous
publication on the uppermost part of the recommendation list.
Consequently, outcome of this recommendation has acme of
recall and precision. Subsequently we can elaborate the
technique for real time application also.
2.
3.
4.
5.
6.
Ricci, F., Rokach, L., Shapira, B., Kantor, P.B. (eds): Recommender
Systems Handbook, pp. 1–35. Springer, Berlin , 2011.
Nikhat Akhtar, Dr. Devendera Agarwal, “A Literature Review of
Empirical Studies of Recommendation Systems” International Journal
of Applied Information Systems (IJAIS), ISSN : 2249-0868, Foundation
of Computer Science FCS, NewYork, USA, Volume 10 – No.2, Pages 6
– 14, Dec 2015, DOI:10.5120/ijais2015451467
Basu, C., H. Hirsh, and W. Cohen. Recommendation as classification:
Using social and content-based information in recommendation. In
Recommender Systems.Papers from 1998 Workshop. Technical Report
WS-98-08.AAAI Press, 1998.
Yusuf Perwej, “An Experiential Study of the Big Data,” International
Transaction of Electrical and Computer Engineers System (ITECES),
USA,ISSN (Print): 2373-1273 ISSN (Online): 2373-1281, Vol. 4, No. 1,
page 14-25, March 2017, DOI:10.12691/iteces-4-1-3.
Nikhat Akhtar, Firoj Parwej, Yusuf Perwej, “A Perusal Of Big Data
Classification And Hadoop Technology,” International Transaction of
Electrical and Computer Engineers System (ITECES), USA, ISSN
(Print): 2373-1273 ISSN (Online): 2373-1281, Vol. 4, No. 1, page
26-38, May 2017, DOI: 10.12691/iteces-4-1-4
Nikhat Akhtar, “A Model- Based Research Material Recommendation
System For Individual Users” for published in the Transactions on
Machine Learning and Artificial Intelligence (TMLAI), Society for
Science and Education, United Kingdom (UK), ISSN 2054-7390,Vol.5,
Issue 2, Pages 1 - 8, March 2017, DOI : 10.14738/tmlai.52.2842
Retrieval Number: E5924018520/2020©BEIESP
DOI:10.35940/ijrte.E5924.018520
14.
16.
17.
18.
19.
20.
21.
22.
REFERENCES
1.
12.
23.
24.
25.
26.
27.
28.
29.
1625
M. Lapata, Automatic Evaluation of Information Ordering: Kendall’s
Tau, Computational Linguistics, vol. 32, no. 4, pp. 471–484, 2006
Q. He, D. Kifer, J. Pei, P. Mitra, and C. L. Giles, “Citation
recommendation without author supervision,” in Proc. of WSDM,
pp.755–764, 2011
Nikhat Akhtar, Dr. Devendera Agarwal, “A Survey of Imperfection of
Existing Recommender Systems for Academic Fraternity”, IOSR
Journal of Computer Engineering (IOSR-JCE), e-ISSN: 2278-0661,
p-ISSN: 2278-8727, Volume 20, Issue 3, Ver. III, PP 08-15, May - June.
2018, DOI:10.5120/ijais2015451467
Beel, J., et al. A comparative analysis of offline and online
evaluations and discussion of research paper recommender system
evaluation. In Proceedings of the international workshop on
reproducibility and replication in recommender systems evaluation.
2013. ACM
D. Yang, B. Wei, J. Wu, Y. Zhang, and L. Zhang. CARES: A
Ranking-Oriented CADAL Recommender System. In Proc. of the 9th
ACM/IEEE Joint Conference on Digital Libraries (JCDL 2009),pages
203–211, 2009
Al Alshaikh, M., G. Uchyigit, and R. Evans, A research paper
recommender system using a dynamic normalized tree of concepts
model for user modeling, 2017
M. D. Ekstrand, P. Kannan, J. A. Stemper, J. T. Butler, J. A.
Konstan,and J. T. Riedl, “Automatically building research reading lists,”
in Proc.ofRecSys, 2010, pp. 159–166.
O. Ku¨c¸u¨ktunc¸, E. Saule, K. Kaya, and U . V. C¸ atalyu¨rek, “Towards
a personalized, scalable, and exploratory academic recommendation
service,” in Proc. of ASONAM, 2013
C. Wang and D. M. Blei, “Collaborative topic modeling for
recommending scientific articles,” in Proceedings of the 17th
ACMSIGKDD International Conference on Knowledge Discovery and
Data Mining, ser. KDD ‟11. New York, NY, USA: ACM, pp. 448–456,
2011
Sholom M. Weiss, Nitin Indurkhya, and Tong Zhang. Fundamentals of
Predictive Text Mining. Springer, 2010.
C. Caragea, A. Silvescu, P. Mitra, and C. L. Giles, “Can‟t see the forest
for the trees?: a citation recommendation system,” in Proc. Of JCDL,
pp. 111–114, 2013
Raamkumar, A.S., S. Foo, and N. Pang, Using author-specified
keywords in building an initial reading list of research papers in
scientific paper retrieval and recommender systems. Information
Processing & Management, 2017. 53(3): p. 577-594
K. El-Arini and C. Guestrin, “Beyond keyword search: discovering
relevant scientific literature,” in Proc of KDD, pp. 439– 447, 2011
Beel, J., et al. Stereotype and Most-Popular Recommendations in the
Digital Library Sowiport. In Everything Changes, Everything Stays the
Same? Understanding Information Spaces. Proceedings of the 15th
International Symposium of Information Science (ISI 2017)-(isi2017),
Humboldt-Universität zu Berlin, 2017
W. Huang, S. Kataria, C. Caragea, P. Mitra, C. L. Giles, and L.
Rokach,“Recommending citations: translating papers into references,”
in Proc.of CIKM, pp. 1910–1914, 2012
Yang chang Zhao. R and data mining: Examples and case studies. In
Yangchang Zhao, editor, R and Data Mining, pages 1–4. Academic
Press, 2013.
X. Liu, Y. Yu, C. Guo, and Y. Sun, “Meta-path-based ranking with
pseudo relevance feedback on heterogeneous graph for citation
recommendation,”in Proc. of CIKM, pp. 121–130, 2014
Tarus, J.K., Z. Niu, G. Mustafa, “Knowledge-based recommendation: a
review of ontology-based recommender systems for e-learning”,
Artificial Intelligence Review, Volume 50, Issue 1, Pages 21-48, 2018
Yusuf Perwej, “An Evaluation of Deep Learning Miniature Concerning
in Soft Computing”, International Journal of Advanced Research in
Computer and Communication Engineering (IJARCCE), ISSN (Online)
: 2278-1021, ISSN (Print): 2319-5940, Vol.4, Issue 2, Pages 10 - 16,
February 2015, DOI: 10.17148/IJARCCE.2015.4203
H. Jingand K. McKeown. Cut and paste based text summarization. In
Prodeedings of NAACL 2000
T. Fujiki, H. Nanba, and M. Okumura, „Automatic acquisition of script
knowledge from a text collection”, EACL 2003 in Budapest, Hungry,
91-94
Mylopoulos, J., “An overview of knowledge representation”, 1981,
ACM press New York, NY, USA
M. Kerber, “Knowledge Representation I”, Introduction to AI
06-08775, 2004/05
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
An Efficient Mining for Recommendation System for Academics
30. Laurence A. F. Park, Marimuthu Palaniswami, and Ramamohanarao
Kotagiri, “Internet document filtering using fourier domain scoring,” in
Principles of Data Mining and Knowledge Discovery, Luc de Raedt and
Arno Siebes, Eds, number 2168 in Lecture Notes in Artificial
Intelligence, pp. 362–373, Springer-Verlag, . September 2001
31. Lin, C.,&Hovy, E. (2001). Neats: A multi document summarizer. In
Proceedings of the document understanding workshop (DUC)
32. I. Ben-Gal, Bayesian Networks, Encyclopedia of Statistics in Quality
and Reliability, 2007
33. Yusuf Perwej, Firoj Parwej, M. M. Mohamed Hassan, Nikhat Akhtar,
“The Internet-of-Things (IoT) Security: A Technological Perspective
and Review”, International Journal of Scientific Research in Computer
Science Engineering and Information Technology (IJSRCSEIT), ISSN :
2456-3307, Volume 5, Issue 1, Pages 462-482, February 2019,
DOI: 10.32628/CSEIT195193
34. Lincy Liptha R, Raja K., G.Tholkappia Arasu,” Enhancing Text
Clustering Using Concept-basedMining Model”, International Journal
of Electronics and Computer Science Engineering, ISSN-2277-1956,
Volume1,Number 2,pp 550-556, 2016
35. Pradnya Randive, Nitin Pise,” Improving Text Clustering Quality by
Concept Mining”, Int. Journal of Engineering Research and
Applications, ISSN : 2248-9622, Vol. 3, Issue 5, pp.1701-1704, Sep-Oct
2013
36. Cem Rıfkı Aydın, Ali Erkan, Tunga Güngör and Hidayet Takçı,”
Dictionary-based concept mining: an application for turkish”,
International Conference on Foundations of Computer Science &
Technology (CST 2014), Zurich, Switzerland, January 2014,
DOI:10.5121/csit.2014.4103
37. K. N. S. S. V. Prasad, S. K. Saritha,” A Survey Paper on Concept
Mining in Text Documents”, International Journal of Computer
Applications (0975 – 8887) Volume 166 – No.11, May 2017
38. K.N.S.S.V. Prasad, S. K. Saritha,” Concept Mining in Text Documents
using Clustering”, International Journal of Computer Applications
(0975 – 8887) Volume 182 – No. 48, April 2019
AUTHORS PROFILE
Nikhat Akhtar, Research Scholar Ph.D (Computer
Science & Engg.) in the Department of Computer Science
& Engineering, Babu Banarasi Das University, Lucknow,
India. She has authored a number of papers in different
journal. Her research interests include Recommender
System, Bayesian Network, Text Categorization and Clustering,
Cryptography, Big Data. He is a member of IEEE.
Retrieval Number: E5924018520/2020©BEIESP
DOI:10.35940/ijrte.E5924.018520
1626
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication