Vellore Institute of Technology (VIT) is a private university located in&nbsp;Tamil Nadu, India. Founded in 1984, as Vellore Engineering College, the institution offers 20 undergraduate, 34 postgraduate, four integrated and four research programs. It has campuses in Vellore, Amravati, Bhopal and Chennai.

VIT is one of the top ranked private universities in India according to NIRF, THE and QS Rankings.&nbsp;Govt. of India has recognized&nbsp;VIT, Vellore as an&nbsp;Institution of Eminence. This has allowed VIT to take independent quality initiatives and move up in world ranking.

&nbsp;

&nbsp;

VIT University

A recommender system (RS) provides assistance for users to filter out items of their interest in the presence of millions of available items. The reason is to find out the likewise user with the assumption that if users have shared similar interest in the past then they may share the same in future. Collaborative filtering (CF) is the widely used recommendation algorithm due to its ease of use but suffers with the problems of sparsity and cold start problem. In this paper, we propose a trust and distrust-based cross domain context aware recommender system in the multi-agent environment which tries to reduce the problem of data sparsity in collaborative-filtering recommender system and improves coverage. Cross Domain Recommender System (CDRS) utilizes data from multiple domains to reduce the problem of sparsity. Moreover, the combination of trust and distrust in recommendation help to improve trustworthiness of generated recommendation. Distrust provides higher accuracy in recommendation by incorporating knowledge about the malicious users. Prototype of the system is developed using JADE and Java technology for the tourism domain consisting of restaurant, hotel, travel places and shopping places as sub-domains. The performance of the proposed trust and distrust-based cross domain recommender system is compared with the traditional approach of recommendation along with the cross domain approach and trust-based cross-domain approach in terms of accuracy and coverage. The results show that the proposed system outperforms in terms of both accuracy and coverage. © 2021 Taylor & Francis.

School of Computer Science and Engineering

Chennai Campus

World University root node

Applied Artificial Intelligence

Trust and Distrust based Cross-domain Recommender System

Group recommender system provides suggestions for a group of users by exploring the choices of individual users of the group. Popularity of group recommender systems is increasing because many activities such as listening to music, watching movies, traveling, etc. are normally performed in groups rather individually. Group recommender systems like personal recommender systems also suffer from cold start and sparsity issues. The cold start and sparsity issues result into inaccurate recommendation computation which degrades the recommendation quality. To handle the cold start and sparsity issues in a Group Recommender System (GRS), this paper proposes to use cross domain approach and introduces Cross Domain Group Recommender System (CDGRS). The recommendations provided by trustworthy and reputed users in the group enhance the acceptance towards the presented recommendations as compared to the other individuals in the group. We have combined the social factors e.g. trust and reputation to get influential user in the group recommendation. A prototype of the system is developed for tourism domain that incorporates four sub-domains i.e. restaurants, hotels, tourist places and shopping places. The performance of CDGRS is compared with GRS. Spearman's Correlation Coefficient, MAE, RMSE, Precision, Recall and F-measure are used to find the accuracy of the generated recommendations. © 2020 - IOS Press and the authors. All rights reserved.

Journal of Intelligent and Fuzzy Systems

Combining trust and reputation as user influence in cross domain group recommender system (CDGRS)

The presence of millions and millions of users and items makes real time filtering a time consuming process in recommender systems (RS). In context aware recommender systems (CARS), choices of users depend on the contextual information as well as available items. The rapid change in interests of a user under different contexts puts extra load on RS. To address this problem, we present a parallel approach for CARS using a multi-agent system which accelerates the processing time drastically using General Purpose Graphic Processing Unit (GPGPU). Contextual pre-filtering and multi-agent environment update the system with the user context. A prototype of the system is developed using JCuda, JADE and Java technologies for tourism domain. The performance of the presented system is compared with the CARS without parallel processing with respect to processing time, scalability as well as precision, recall and F-measure. The results show a significant speedup for presented system over non-parallel CARS. Copyright © 2019 Inderscience Enterprises Ltd.

International Journal of Computational Science and Engineering

Parallel context-aware multi-agent tourism recommender system

User profile modelling for the domain of tourism is different compared with most of the other domains, such as books or movies. The structure of a tourist product is more complex than a movie or a book. Moreover, the frequency of activities and ratings in tourism domain is also smaller than the other domains. To address these challenges, this study proposes a trust and reputation-based collaborative filtering (TRbCF) algorithm. It augments a notion of dynamic trust between users and reputation of items to existing collaborative approach for generating relevant recommendations. A multi-agent recommender system for e-tourism (MARST) for recommending tourism services using TRbCF algorithm is designed and a prototype is developed. TRbCF also helps to handle new user cold-start problem. The developed system is capable to generate recommendations for hotels, places to visit and restaurants at a single place whereas most of the existing recommender systems focus on one service at a time. © 2018 Inderscience Enterprises Ltd.

Trust and reputation-based multi-agent recommender system

Recommender systems (RS) suffer from cold start and data sparsity problem. Researchers have proposed various solutions to this problem in which cross domain recommendation is an effective approach. Cross domain recommender system (CDRS) utilizes user data from multiple domains to generate prediction for the target user. This paper proposes a proactive cross domain recommender system. This paper also introduces a parallel approach in cross domain recommendation using general purpose graphic processing unit (GPGPU). This will help to accelerate the computation in the multi-agent environment as data processing in multiple domains takes significant amount of time. A prototype of the system is developed in tourism domain using Cuda, JCuda, Java, Android studio and Jade. The system uses four domains which is restaurant, tourist places, shopping places and hotels. The performance of the parallel CDRS system is compared with non-parallel CDRS in terms of their processing speed. Also the system is compared to the normal Collaborative Filtering approach to measure accuracy of the proposed system using MAE as well as precision, recall and F-measure. The results show a significant speedup for the presented system over non-parallel system. © 2018 - IOS Press and the authors. All rights reserved.

Parallel proactive cross domain context aware recommender system

Proactive recommender systems are intelligent systems that provide (i.e. push) pertinent recommendations to the users based on their current tasks or interests. The recommendation algorithms employed in these systems usually compute similarity score or build up a model offline using training data for producing online recommendations. As training in proactive recommender systems when the availability of items changes often and rapidly is very time consuming, existing recommendation algorithms are less effective in such application domains. To address this problem, we present a proactive recommender system that generates real time recommendations using the proposed Extreme Learning Machine based Imputation-Boosted Collaborative Filtering (ELMICF) algorithm. Extreme Learning Machine (ELM) is a machine learning algorithm which considerably reduces the time required for training the system due to the very fast learning process of ELM. It has been used in literature for numerous classification, generalization and prediction applications. ELMICF first employs an imputation technique to handle data sparseness in the input rating matrix and then uses the ELM as a classifier to predict the novel ratings. A prototype of the system has been implemented for restaurant recommendations to show the feasibility of our proposed approach. The performance of ELMICF is compared with MLP/ANN and naïve based classification techniques using normalized Discounted Cumulative Gain (nDCG), average precision, training time and mean prediction time metrics. © 2016 IEEE.

2016 International Conference on Advances in Computing, Communications and Informatics, ICACCI 2016

ELM based imputation-boosted proactive recommender systems

Recommender Systems (RS) are intelligent applications which predict the preferences for users from their current interests. Collaborative Filtering (CF) is a widely used Recommendation technique. On large scale data, collaborative filtering approach is very time consuming and hence parallel processing can be useful for accelerating the task of recommendation. Context Aware Recommender Systems (CARS) help us to generate precise and relevant recommendations by using specific context of the user. This paper presents Parallel Context Aware CF Recommender Systems using JCuda (PCARS) that involves Graphic Processing Unit (GPU). The proposed algorithm works in two phases: offline and online processing. Two kernels are identified and processed for offline processing which reduce the processing time drastically. Online processing involves with the prediction calculation for the items which is not yet seen by target user and can be recommended to them. Offloading the prediction calculation to GPU will help to increase the overall system performance significantly. A prototype of the proposed system is developed using JCuda, CUDA and Java technologies for the restaurant domain. The performance of the proposed system PCARS is compared to the Collaborative Filtering Recommender Systems (CFRS) without parallel processing using contextual pre-filtering and context post-filtering in terms of processing time. Experimental results demonstrated the tremendous speedup of the system over a single-core CPU based system. © 2016 IEEE.

Parallel context aware recommender system using GPU and JCuda

In this paper, a novel user interest expansion approach for generating recommendations is proposed. The approach utilizes interest of user as well as semantic relatedness between the items along with context to generate recommendations. Ontologies are used to represent domain knowledge. Spreading activation technique uses the relatedness between concepts of user interest in domain ontology to expand user interest which results in generation of diverse recommendations. A prototype of the system has been designed and developed using various Java technologies for Restaurant domain and its performance is evaluated using precision, recall, F1 and diversity metrics. Performance of the proposed Context Aware Recommender Systemwith Expansion of user interest (ECARS) is compared with Context Aware Recommender System (CARS) and Content Based Recommender System (CB) for generating recommendations. © 2015 IEEE.

2015 International Conference on Advances in Computing, Communications and Informatics, ICACCI 2015

User interest expansion using spreading activation for generating recommendations

This paper presents a Multi-Agent Recommender system for e-Tourism (MARST) for recommending tourism services to the users. This system uses Reputation based Collaborative Filtering (RbCF) algorithm that augments reputation to existing Collaborative approach for generating relevant recommendations and to handle cold-start new user problem in tourism domain. The structure of a tourist product is more complex than a book or a movie and hence user profile modeling for these systems is much harder than most of other applications domains like books or movies. Moreover the frequency of activities and rating in tourism domain is also much smaller than in most of the other domains. This increases the complexity in designing and development of Recommender Systems in tourism domain. An attempt has been made in this paper to generate relevant services for a user in tourism domain using reputation based collaborative filtering. Most of the existing Recommender systems focus on one service at a time, whereas the proposed system incorporates three services (hotels, places to visit and restaurants) at a single place to ease the searching of information at one place only. The prototype of MARST has been designed and developed using various JAVA technologies and its performance was evaluated using precision, recall and F1 metrics. © 2014 Springer International Publishing.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

MARST: Multi-Agent Recommender System for e-Tourism using reputation based collaborative filtering

Recommender systems assist users to narrow down the choices from the plethora of options available to them, by recommending options most suitable to their profile. Traditional Collaborative Filtering (CF) technique suffers from the problem of popularity bias as a result of which the items recommended lack novelty in them. The need today is, to incorporate novel items in the recommended list of items, as popular items are apparent and lack novelty. This paper proposes a new Novelty Score (NS) metric based on item frequency and inverse user frequency for unseen items. The paper also proposes Modified Collaborative Filtering Approach for Novel Recommendations (MCFNR)which utilizes this metric to generate novel recommendations. The target user is categorized as occasional or persistent in this work. Collaborative Filtering is used to generate recommendations for occasional users whereas MCFNR is used to generate recommendations for persistent users as popularity bias affect such users the most MCFNR eradicates the problem of popularity bias while keeping the relevance by introducing the concept of novelty in the recommendations. The prototype of Recommender System is developed and tested with Books data set as a case study. Results are evaluated for CF and MCFNR for persistent users using precision, recall and F-measure. © 2014 IEEE.