296 IISWS: Integrative Intelligent System for a Multi-Domain Diversified Semantic Search Gerard Deepak a, Santhanavijayan A a aDepartment of Computer Science and Engineering, National Institute of Technology, Tiruchirappalli, India Abstract The information density is steeply rising over the World Wide Web and there is an urgent need of an integrative intelligent approach for facilitating recommendations from the Web owing to the data density of the Web. Knowledge-Driven Web-based recommendation systems are required to reduce the cognitive gap between the user query and the recommendable contents. An Integrative Intelligent approach for knowledge-centric recommendations from the Web has been proposed with the unification of SPARQL Endpoints from several heterogeneous sources through a domain indexing service that yields reasonable dynamic knowledge into the recommendation framework. The approach integrates classification, Synonymization, Intertwining with RDF, and a cognizable semantic similarity model for recommendation of queries and thereby the user relevant web pages with the diversification of search results. The proposed IISWS furnishes an overall accuracy of 91.84% with a small FDR of 0.1 for the WebKB Corpus dataset. A Normalized Discounted Cumulative Gain of 0.91 has been achieved by the IISWS. Keywords Content Based Filtering, Hash Table, Knowledge Modeling, Ontologies, Semantic Latent Analysis, Semantic Web 1. Introduction These systems do not facilitate the diversification of search results as there is a sparsity of real-world The World Wide Web is the largest storehouse of knowledge that is being instilled into the system. Owing Information. The Web 2.0 is revolutionizing into the to the evolution of the existing Web 2.0 into a Semantic Web which is the “Web of Data” with very perspicuous Semantic Web, a large amount of data high data density and cohesiveness. Information available on the current structure of the Web and to Extraction from the current configuration of the World reason out, a strong semantic approach with can Wide Web is not just a laborious task but also involves a transform the existing data into knowledge or which can lot of effort to link the user queries with the contents in incorporate knowledge for the recommendation. A the World Wide Web. Even if there is a proper learning knowledge-centric web search is the most amicable in mechanism incorporated, most web-based search the current situation where there is an evident transition algorithms tend to lag owing to the density of the from the conventional Web to the Semantic Web. There information that looks quite similar. Most of the Web- are also a large number of domain-centric knowledge based recommendation system are either only query- bases and RDF stores that are authorized and a few of centered or user-centered. The query-centric web page them are collective revised and newer versions are recommendation systems focus only on the relevance of released progressively. It is strategic when the the web pages to the query that is being entered into the knowledge that has been dynamically integrated based system. However, when the web page recommendation on Intertwining or Interlinking of the RDF structures system is user-centric, the focus is mainly on the from the World Wide Web to the semantic inference personalization and satisfying the needs of the user. Most algorithm for facilitating knowledge-driven web search of these queries centric and user-centric systems are quite and yield diversified and yet contents that are relevant to efficient and have tackled the problems of synonymy and the query and also satisfy the user needs, based on the polysemy. diversification of contents. Motivation: Knowledge-Driven paradigms ISIC’21: International Semantic Intelligence Conference, February with low computational complexity is the need of the 25-27, 2021, Delhi, India hour in Web-based Recommendation Systems to cater to :gerard.deepak.christuni@gmail.com(Gerard Deepak); vijayana@nitt.edu (Santhanavijayan A) the demanding informational needs of users. Also, the ORCID: 0000-0003-0466-2143 (Gerard Deepak) Web-based Recommendation Systems needs to be quite ©️ 2021 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). lightweight by transforming the traditional learning CEUR Workshop Proceedings (CEUR-WS.org) paradigm into reasoning based inferential strategy for recommending query relevant and user-relevant information from the Web. 1 297 Contributions: A dynamic knowledge-centric domain. In [8-11] Deepak et al., have incorporated model for recommending queries and web pages based several intelligent semantics paradigms in combination on the standards and constructs of Web 3.0 has been with ontologies for recommending web contents in web proposed. The approach integrates an initial search and personalization. Deepak et al., [12] have classification model that carefully selects the top-25% of proposed the Differential Semantic Algorithm (DSA) the most relevant results to cognitively enrich an RDF that uses differential threshold on semantic similarity driven real-world knowledge from a series of SPARQL algorithms for facilitating personalized web search. Endpoints based on the domain of relevance of the Query. The strategy blends a WordNet 3.0 based Elshaweesh et al., [13] have personalized and Synonymization as well as Intertwining Real-World recommended web pages based on user profile analysis, Knowledge through indexing services and SPARQL latent semantic analysis, and the browsing behavior of Endpoints that are quite heterogeneous. Also, an the user with semantic knowledge via the Ontologies. inferential cognizable semantic similarity by fusing the Sumathi et al., [14] have proposed the IFWIAR for query Jaccard and the NPMI model is used to inferentially recommendation based on the usage of a domain specific select the recommendable entities. Experimentations are ontology with improved weighted fuzzy iterative rule- conducted on the WebKB Corpus and an overall based ontology processing. Though they have achieved accuracy of 0.91 with a reasonably small FDR of 0.1 has results that are appreciable, the knowledge sparsity is been achieved. The proposed IISWS is validated by two visible as it is highly domain specific and integration of baseline models and knowledge-centric variations to real-world knowledge isn’t visible. Omar et al., [15] have prove the efficacy of derived knowledge and its role in introduced a personalized approach for integrating web page recommendation. domain knowledge and user profile-based ontology for transportation domain using WordNet API and semantic Organization: The rest of paper is organized as conceptual knowledge. follows. The Literature Survey is detailed in Section 2. The Proposed Architecture is described in Section 3. The Implementation in represented is in Section 4. The 3. Proposed Architecture Performance Evaluation are portrayed in Section 5. The Conclusions are formulated in Section 6. Figure 1 judiciously describes the architecture of the proposed which intelligently integrates the phases of initial classification and integration of real-world 2. Related Literature knowledge to facilitate web page recommendation. The proposed IISWS does not make use of any static Soto et al., [1] have put forth a semantic search engine which is domain-specific for biomedical abstracts that ontologies or knowledge resources, instead a dynamic update auxiliary knowledge from PubMed on a daily ontology modeling scheme for upper ontology derivation basis. The search engine uses a NER approach for has been imbibed into the approach using existing real- biomedical entities and a context-sensitive Acronym world collaborative cognitive knowledge to ensure that Resolution for concept recognition. Xie et al., [2] have recommendations are not restricted to the conceptual recommended web pages based on two-fold clustering level but also yield enough individuals when there is a where the relationship between the user behavior and conflict of interest, specifically when homonymous topics are correlated. Bhavithra et al., [3] have formulated an approach for case-based reasoning terms are identified by the Web Search System. The core focusing on clustering as a paradigm which is based on principle behind the IISWS is to overcome the Sparsity weighted association rule mining for recommending web problem in Web Search and strategically overcome the pages. Singh et al., [4] have imbibed sequential rules problems of synonymy, polysemy, and homonymy. partially ordered web page recommendation which Also, the design of IISWS is in away such that predicts the future interests of the users’ web page Serendipity problem is also overcome by ensuring the information. Katarya et al., [5] have incorporated the fuzzy c-means clustering technique for the user query is subject to Query Pre-processing such as the recommendation of web pages that makes use of users’ Tokenization, Stop Word navigation details of web pages. Removal, and Lemmatization. On the pre-processing of Ontologies have been used in several possible the query terms, a query word set is furnished. In order combinations in several Web Search paradigms. Ali et to enhance the probability of integration of likewise al., [6] have used Fuzzy Ontologies in combination with entities, the Query Word Set is subject to SVM for web content classification. Thanapalasingam et Synonymization where several Synsets for a specific al., [7] have recommended editorial products based on query word is anchored with the Query word. The the domain ontology model. They have depicted the usefulness of domain ontology and its impact in Synonymization was done based on the integration with recommendation specific to the computer science a WordNet 3.0 synset generator, and several 2 298 relationships like the Homonym terms, and terms with polysemous contents has been identified at this stage. However, the synonyms generated with polysemous and homonym terms are reserved for future knowledge enrichment during the Intertwining phase. Figure 1: Architecture of the Proposed IISWS The initial query words generated are subject to initial owing to the reason that classification need not be level classification of the labelled dataset that has been emphasized for being highly accurate, and the cleaned and pre-processed. The classification is done methodology requires approximate classification for based on a traditional Naïve Bayes Classifier by fixing future reasoning and inferencing during the class labels as the pre-processed query word set. The recommendation. For each of the classes, top 25% of reason for using a conventional classifier is mainly classification results under each label, i.e., the query owing to the reason that classification need not be word is taken into consideration and is transformed into emphasized for being highly accurate, and the its equivalent RDF structure. Once the classification methodology requires approximate classification for results in the top 25% is transformed into RDF, the future reasoning and inferencing during process of Intertwining is performed where the RDF is recommendation. For each of the classes, top 25% of linked with real-world SPARQL Endpoint server to classification results under each label, i.e., the query associate the real-world entities along with their word is taken into consideration and is transformed into equivalent class labels and formulate a linked RDF its equivalent RDF structure. Once the classification structure with dynamic aggregation of real-world results in the top 25% is transformed into RDF, the ontological knowledge in the form of RDF data. process of Intertwining is performed where the RDF is linked with real-world SPARQL Endpoint server to The SPARQL Endpoint is linked to Wikidata, DBPedia, associate the real-world entities along with their LinedCT, voiD Data, and UniProt based on the query equivalent class labels and formulate a linked RDF centric domain relevant identification of entities. structure with dynamic aggregation of real-world However, each of these data stores and their domain ontological knowledge in the form of RDF data. availability is identified with the help of an interim repository that has been indexed with the domain and The initial query words generated are subject to initial sub-domain terminologies. Once the Intertwining has level classification of the labelled dataset that has been been successfully staged, there is synonym induction that cleaned and pre-processed. The classification is done has been computed previously in order to further enrich based on a traditional Naïve Bayes Classifier by fixing the density of query centered domain knowledge based the class labels as the pre-processed query word set. The on dynamic Ontology Generation through RDF reason for using a conventional classifier is mainly Intertwining. The RDF Intertwining generates extensive 3 299 Cognitive Knowledge for formulation of queries and queries and further choose the relevant web pages. The recommending query words based on Bigram and Jaccard Similarity is depicted by Equation (1) and the Trigram integration. The methodology requires an NPMI is depicted in Equation (2). The Normalized intelligent recommendation strategy for regulating the pointwise mutual information is dependent on the query formulation which is done based on the pointwise mutual information measure portrayed by computation of the standard Jaccard Similarity and the Equation (3). Equation (4) illustrates the intersection Normalized Pointwise Mutual Information Computation incidence of both the Jaccard Similarity and the Scheme, wherein the top 15% integration of common Normalized Pointwise Mutual Information measure. elements from either of the two schemes has been | 𝑆 ∩ 𝑇| formulated. The recommendation strategy has been 𝐽𝑎𝑐𝑐𝑎𝑟𝑑(𝑆, 𝑇) (1) |𝑆| + |𝑇| − |𝑆 ∩ 𝑇| depicted in the Intelligent Inference Algorithm that has been depicted in the further sections. The recommendation results are re-arranged in the declining order of the values of semantic similarity, and further 𝑝𝑚𝑖(𝑥; 𝑦) 𝑛𝑝𝑚𝑖(𝑥; 𝑦) = (2) yielded to the user. The results have a high degree of − log[𝑝(𝑥, 𝑦)] diversity as the user has a lot of choices to select because of the strategies followed in the proposed approach. Further, if the user is dissatisfied with the yielded results, 𝑝𝑚𝑖(𝑥; 𝑦) = ℎ(𝑥) + ℎ(𝑦) − ℎ(𝑥, 𝑦) (3) then the query words are re-substituted by the current user click sub-topics in the recommendation, and thereby driving the recommendation to a much-focused topic and SemanticSimilarity= |Jaccard | ∩ | NPMI | (4) thereby altering the feasible recommendable terms until the user is satisfied. The semantic similarity is computed using the intersection of the Jaccard Similarity and the Normalized Pointwise Mutual Information (NPMI). The 4. Implementation Jaccard Similarity is computed with a threshold of 0.75 The implementation has been achieved using the Python while the NPMI is considered between 0 and 1 without NLTK library and the OntoSpy library. However, the yielding to its negative values. The threshold of 0.5 is SPARQL Wrapper interface was modeled with an agent considered for the NPMI. Further the intersection of the such that the SPARQL endpoints can be queried for datapoints between the individual threshold of the Intertwining. However, a customized domain index Jaccard Similarity and the NPMI is taken into repository has been used for mapping and indexing the consideration for recommending the terms to formulate terms. Table 1: Proposed Integrative Intelligent Algorithm (IIA) for Web Page Recommendation Input: Initially Classified Dataset based on the Query Words as Labels, Query Word Set w, Real World Knowledge Bases, Output: Recommended Expanded Queries and their corresponding Web Pages Begin Step 1: while (Qw.next()!=NULL) for each Qw as Label HashSet R’ ←Select Top 25% and Convert into RDF end for end while Step 2: for each in R’ 2.1Select the domain index for R’.current() from a Thesauri 2.2 Based on Index Domain trigger SPARQLEndpoint 2.3 Generate Intertwined Knowledge as Tree Iw end for Step 3: Generate SynSets of Qw as Qsyn 4 300 Step 4: while (Qsyn.next()!=NULL) Tree IEnriched←Repeat Step 2 for Qsyn.current() end while Step 5: while (Qw.next()!=NULL) Set J←JaccardSim(Qw, IEnriched.pos()) Set N←NPMI ((Qw, IEnriched.pos()) return R←J∩N end while Step 6: Choose any two levels from R as either super concepts or sub concepts or From its neighbor and recommend by applying Bigram & Trigram and Recommend to the user. Step 7: Repeat Step 5 by matching elements of user-click and by not visiting a single node more than once. Step 8: If there exist user clicks, then substitute the maximum topic of user click as Qw and begin from Step 1 until there are no further user clicks. End The proposed Integrative Intelligent Algorithm for Web 5. Performance Evaluation Page Recommendation is depicted in Table 1 which takes input as the Classified Datasets based on the Query The experimentations for the proposed IISWS has been Labels, Query Word Set, and SPARQL Endpoints via a carried out for the WebKB Corpus with 492 benchmark domain indexing service for the Real-World Knowledge test queries. 72 users were given 40 queries each to give Stores. The Algorithm furnished the recommended their top 10 recommendations in terms of both query expanded queries at first, and further their corresponding categories and individual web pages. However, each user web pages. The algorithm selects the top 25% of the got different variations of the same query, and finally for classified data from each of the classes with query words each query, top 10 frequently occurring webpage and the as a label and is transformed into its equivalent RDF. queries were considered as ground truth for Further, the domain index for the class labels and experimentation. randomized classes is selected, and based on the domain index a set of SPARQL Endpoints are selected for The Precision, Recall, Accuracy, F-Measure, and False including relevant real-World Knowledge for Discovery Rate (FDR) were used as standard metrics for Intertwining the RDF and formulate a Knowledge Tree, evaluating the performance of the proposed IISWS. which is further enriched based on the initial set of Also, nDCG (Normalized Discounted Cumulative Gain) synonyms generated. The generated synonyms are also was used to measure the diversity of the results yielded. intertwined into real-world knowledge. The selection of Equations (5), (6), (7), (8), and (9) depict the Precision, recommendable query entities is realized on the basis of Recall, Accuracy, F-Measure, and FDR. Equations (10) the semantic similarity computation and the and (11) represent the Normalized Discounted recommendation of queries is done by bigram and Cumulative Gain and the Discounted Cumulative Gain trigram formulation based on user query clicks. The web respectively. pages are selectively displayed based on the user-click of Retrieved∩Relevant the recommended queries. The process is continued until Precision= (5) Retrieved each node of the enriched knowledge tree is visited or until the user has no clicks recorded. If all the nodes of Recall= Retrieved∩Relevant (6) Relevant the current knowledge tree have been visited, then the last few user clicks are once again considered as query Precision+Recall words to facilitate newer recommendations until the user Accuracy= (7) 2 is satisfied, i.e., till no further user clicks are recorded. 5 301 2×Precision×Recall models and other knowledge level variations in the same F-Measure= (8) Precision+Recall environment as of the proposed IISWS. FDR=1-PPV (9) It is evident from Table 2 that the proposed IISWS furnishes an average precision, recall, and accuracy of DCGα nDCG = (10) 90.21%, 93.47%, and 91.84% respectively. IISWS IDCGα furnished a very low FDR of 0.1 with an nDCG of 0.91 ∝ which is better performing than the baseline models and 𝑅𝑒𝑙𝑖 its variants. The IISWS yields a 13.6% higher accuracy 𝐷𝐶𝐺 = ∑ (11) than DSA and 6.38% higher accuracy than the IFWIAR. 𝑙𝑜𝑔(𝑖 + 1) 𝑖=1 The reason for a better performance of the IISWS than Since Query Recommendation is followed by the web the DSA is due to the reason that IISWS has an efficient page recommendation in the proposed IISWS, the semantic model and heterogenous integrative knowledge performance measures were computed for the final web form several real-world knowledge resources through pages that have been furnished by the proposed system. SPARQL Endpoints. The average performance evaluation for the WebKB Corpus has been depicted in Table 2 where the proposed The DSA has a strong strategy of regulating the IISWS and has been benchmarked with several other recommendation by a combination of semantic similarity models and variants of knowledge. Also, a variation measures with varied thresholds, however, the DSA only without specific knowledge model has been depicted. focuses on personalization, and does not use any external The IISWS is also baselined with the Differential static knowledge or dynamic inferential knowledge. Semantic Algorithm (DSA) [12] that uses a semantic IFWIAR encompasses Ontologies with fuzzy weighted strategy for recommending web pages and IFWIAR [14] iterative association rule with static domain ontologies. that uses ontologies for yielding the web pages. The However, owing to the sparsity of domain ontologies experimentations are conducted for every baseline supplied into the IFWIAR, there is a predominant lag in its performance. Table :2 Performance Analysis of the Proposed IISWS Framework Model/ Variants Precision Recall Accuracy FDR nDCG % % % DSA 76.14 79.74 78.24 0.24 0.76 IFWIAR 84.17 86.74 85.46 0.16 0.86 Static Ontological Model 84.12 87.74 85.93 0.16 0.85 Fuzzy Ontologies 83.14 85.74 84.44 0.17 0.84 Without any Knowledge Model 79.47 81.14 76.98 0.21 0.74 Single Source SPARQL Endpoint 82.14 84.32 83.23 0.18 0.82 Proposed IISWS 90.21 93.47 91.84 0.1 0.91 6 302 Recommendations for the proposed IISWS framework and its variants of the baseline models is depicted in Figure 2 from where it is clearly evident that the F- Measure Distribution of the proposed IISWS is much higher when compared to the other variants in terms of knowledge and Ontological Models and the baseline systems namely the DSA and the IFWIAR. The proposed IISWS framework not just performs based on the relevance of the finally recommended results but also focuses on the diversification of results based on the integration of knowledge from varied sources. The diversity of results is quantitatively measured using the nDCG. It is evident from Figure 3 that the nDCG is the highest for the proposed IISWS framework and it measures to 0.91. However, the DSA and IWFIAR have a nDCG of 0.76 and 0.86 respectively. Even the variants Figure 2: F-Measure Vs Number of of knowledge like that of Static Ontology Models and the Recommendations Fuzzy Ontologies have a nDGC of 0.85 and 0.84 respectively. The Single Source SPARQL Endpoint has Since the strategic amalgamation of auxiliary knowledge a nDGC of 0.82. The absence of any knowledge model along with a strong inferential semantic similarity model has the lowest nDGC of 0.74. The reason for the is a requisite in the viscinity of the linked semantic data diversification is owing to the rich amount of entities that for achieving a higher performance, variations in regard are dynamically generated from a system of knowledge to the knowledge models were demonstrated to show the stores and repositories based on the nature of domain effectiveness of the IISWS model which has been put through SPARQL Endpoints. Since there is diversity and forth. The replacement of the heterogeneous dynamic the density in the supplied auxiliary knowledge, the knowledge model with a single source SPARQL proposed IISWS outperforms the baseline approaches Endpoint decreased the Accuracy of the IISWS by and the other variants of the same algorithm in terms of 8.61%. Further, the use of Fuzzy Ontologies and the the knowledge models. Static Ontologies, decreased the Accuracy of IISWS by 7.4% and 5.91% respectively. When a Single Source 6. Conclusions SPARQL Endpoint is incorporated, there is no sufficient knowledge amalgamation for the current query. An Intelligent Integrative approach for Knowledge However, when the Static Ontological Model is used, Centric web page recommendation has been proposed. there is sufficient auxiliary knowledge but modeling the The IISWS system initially recommends diversified same is cumbersome. The Fuzzy Ontologies are queries and then the web pages on the basis of the comparatively less domain-specific than that of the Static knowledge rendered by heterogenous SPARQL Ontologies thereby exhibiting a lesser accuracy than that Endpoints. The IISWS intelligently encompasses an of the Static Ontologies. Finally, in the absence of a initial classification of the dataset to choose the top 25% standard knowledge model in the environment of the of each of the classification, which is further converted inferential recommendation model of the IISWS, the into RDF that is further interlinked with real-world accuracy decreased by 14.86% which clearly ensures that knowledge stores through the Multi-source SPARQL amalgamation of real-world dense cognitive knowledge Endpoints based on selective domain integration. The reduces the strategic gap between the user-query and the dynamic generation of the query relevant entities, finally recommended items. synonymization, and an efficient semantic similarity- based recommendation of the web pages makes IISWS The accuracy directly correlates with the Precision, as the most desirable web page recommendation system. Recall, and F-Measure and is inversely proportional to Moreover, the enrichment of knowledge into the system the FDR. With the increase in the Accuracy, there is a ensures diversified and yet query relevant significant decrease of the FDR. The IISWS has the recommendations. An overall accuracy of 0.91 with a lowest FDR of 0.1 when compared to that of DSA and reasonably small FDR of 0.1 and a nDCG of 0.91 has IFWIAR and all the other variants. The distribution of the percentage of F-Measure Vs the Number of 7 303 been attained by the proposed IISWS Framework for the WebKB Corpus dataset. [9]Deepak, G., & Priyadarshini, J. S. (2018). Personalized and Enhanced Hybridized Semantic References Algorithm for web image retrieval incorporating ontology classification, strategic query expansion, and [1] Soto, Axel J., Piotr Przybyła, and Sophia Ananiadou. content-based analysis. Computers & Electrical "Thalia: semantic search engine for biomedical Engineering, 72, 14-25. abstracts." Bioinformatics 35.10 (2019): 1799-1801. [10] Bhavithra, J., and A. Saradha. "Personalized web [2] Xie, Xianfen, and Binhui Wang. "Web page page recommendation using case-based clustering and recommendation via twofold clustering: considering user weighted association rule mining." Cluster Computing behavior and topic relation." Neural Computing and 22.3 (2019): 6991-7002. Applications 29.1 (2018): 235-243. [11] Deepak, G., & Santhanavijayan, A. (2020). [3] Bhavithra, J., & Saradha, A. (2019). Personalized OntoBstFit: A Best-Fit Occurrence Estimation strategy web page recommendation using case-based clustering for RDF driven faceted semantic search. Computer and weighted association rule mining. Cluster Communications, 160, 284-298. Computing, 22(3), 6991-7002 [12] Deepak, Gerard, J. Sheeba Priyadarshini, and MS [4] Singh, H., Kaur, M., & Kaur, P. (2017). Web page Hareesh Babu. "A differential semantic algorithm for recommendation system based on partially ordered query relevant web page recommendation." 2016 IEEE sequential rules. Journal of Intelligent & Fuzzy Systems, International Conference on Advances in Computer, 32(4), 3009-3015. ICACA, Coimbatore, India. [5] Katarya, Rahul, and Om Prakash Verma. "An [13] ElShaweesh O, Hussain FK, Lu H, Al-Hassan M, effective web page recommender system with fuzzy c- Kharazmi S. Personalized Web Search Based on mean clustering." Multimedia Tools and Applications Ontological User Profile in Transportation Domain. 76.20 (2017): 21481-21496. InInternational Conference on Neural Information [6] Ali, F., Khan, P., Riaz, K., Kwak, D., Abuhmed, T., Processing 2017 Nov 14 (pp. 239-248). Springer, Cham. Park, D., & Kwak, K. S. (2017). A fuzzy ontology and [14] Sumathi G, Akilandeswari J. Improved fuzzy SVM–based Web content classification system. IEEE weighted‐iterative association rule-based ontology Access, 5, 25781-25797. postprocessing in data mining for query recommendation [7] Thanapalasingam T, Osborne F, Birukou A, Motta E. applications. Computational Intelligence. 2020 Ontology-based recommendation of editorial products. May;36(2):773-82. InInternational Semantic Web Conference 2018 Oct 8 [15]Al-Hassan, Malak, and Sadegh Kharazmi. (pp. 341-358). Springer, Cham. "Personalized Web Search Based on Ontological User [8] Deepak, G., Ahmed, A., & Skanda, B. (2019). An intelligent inventive system for personalised webpage Profile in Transportation Domain." In Neural recommendation based on ontology semantics. In the Information Processing: 24th International Conference, International Journal of Intelligent Systems ICONIP 2017, Guangzhou.,China, November14–18, 20 Technologies and Applications, 18(1-2), 115-132. p. 239, Proceedings by Springer, 2017. 8