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