Gunwant Mankar

ABSTRACT:

We propose TrustSVD, a trust-based matrix factorization technique for recommendations. Trust SVD integrates multiple information sources into the recommendation model in order to reduce the data sparsity and cold start problems and their degradation of recommendation performance. An analysis of social trust data from four real-world data sets suggests that not only the explicit but also the implicit influence of both ratings and trust should be taken into consideration in a recommendation model. TrustSVD therefore builds on top of a state-of-the-art recommendation algorithm, SVD++ (which uses the explicit and implicit influence of rated items), by further incorporating both the explicit and implicit influence of trusted and trusting users on the prediction of items for an active user.The proposed technique is the first to extend SVD++ with social trust information. Experimental results on the four data sets demonstrate that Trust SVD achieves better accuracy than other ten counterparts recommendation techniques.

PROJECT OUTPUT VIDEO:

EXISTING SYSTEM:

• Many approaches have been proposed in this field, including both memory- and model-based methods.
• Golbeck proposes a Tidal Trust approach to aggregate the ratings of trusted neighbors for a rating prediction, where trust is computed in a breadth-first manner.
• Guo et al. complement a user’s rating profile by merging those of trusted users through which better recommendations canbe generated, and the cold start and data sparsity problem scan be better handled. However, memory-based approaches have difficulty in adapting to large-scale data sets, and are often time-consuming to search candidate neighbors in a large user space.
• Zhu et al. propose a graph Laplacian regularizer to capture the potentially social relationships among users,and form the social recommendation problem as a low ranks emi-definite problem. However, empirical evaluation indicates that very marginal improvements are obtained incomparison with the RSTE model.
• Yang etal. propose a hybrid method TrustMF that combines both a truster model and a trustee model from the perspectives of trusters and trustees, that is, both the users who trust the active user and those who are trusted by the user will influence the user’s ratings on unknown items.

DISADVANTAGES OF EXISTING SYSTEM:

• Existing trust-based models may not work well if there exists only trust-alike relationships.
• These observations could other kinds of recommendation problems.
• Existing trust based models consider only the explicit influence of ratings.
• The utility of ratings is not well exploited.
• Existing trust-based models do not consider the explicit and implicit influence of trusts imultaneously.

PROPOSED SYSTEM:

• We propose a novel trust-based recommendation model regularized with user trust and item ratings, termed as Trust SVD.
• Our approach builds on top of a stateof-the-art model SVD++ through which both the explicit and implicit influence of user-item ratings are involved to generate predictions. In addition, we further consider the influence of trust users (including trustees and trusters) on the rating prediction for an active user.
• This ensures that user specific vectors can be learned from their trust information even if a few or no ratings are given. In this way, the concerned issues can be better alleviated.
• Therefore, both explicit and implicit influences of item ratings and user trust have been considered in our model, indicating its novelty. In addition, a weighted-regularization technique is used to help avoid over-fitting for model learning.
• The experimental results on the data sets demonstrate that our approach works significantly better than other trust-based counterparts as well as other ratings-only high-performing models (ten approaches in total) in terms of predictive accuracy, and is more capable of coping with the cold-start situations.
• There are two main recommendation tasks in recommender systems, namely item recommendation and rating prediction. Most algorithmic approaches are only (or best) designed for either one of the recommendations tasks, and our work focus on the rating prediction task.

ADVANTAGES OF PROPOSED SYSTEM:

• Our first contribution is to conduct an empirical trust analysis and observe that trust and ratings can complement to each other, and that users may be strongly or weakly correlated with each other according to different types of social relationships.
• These observations motivate us to consider both explicit and implicit influence of ratings and trust into our trust-based model.
• Potentially, these observations could be also beneficial for solving other kinds of recommendation problems, e.g., top-N item recommendation.

MODULES:

• System Construction
• Rating Prediction
• Item Recommendation
• A Trust-Based Recommendation Model

MODULES DESCSRIPTION:

System Construction

• In the first module, we construct social rating based system construction module for the implementation of our proposed model. In this module we design to havewidely used to provide users with high-quality personalized recommendations from a large volume of choices. Robust and accurate recommendations are important in e-commerce operations (e.g., navigating product offerings, personalization,improving customer satisfaction), and in marketing (e.g.,tailored advertising, segmentation, cross-selling). In this system we focus on user-item ratings,ItemRating Prediction, user can recommend a item to their friends.
• In the first module, we construct social rating based system construction module for the implementation of our proposed model. In this module we design to havewidely used to provide users with high-quality personalized recommendations from a large volume of choices. Robust and accurate recommendations are important in e-commerce operations (e.g., navigating product offerings, personalization,improving customer satisfaction), and in marketing (e.g.,tailored advertising, segmentation, cross-selling). In this system we focus on user-item ratings,ItemRating Prediction, user can recommend a item to their friends.
• Where users can also share post with others. The user can able to search the other user profiles and public posts. In this module users can also accept and send friend requests.
• With all the basic feature of Online Social Networking System modules is build up in the initial module, to prove and evaluate our system features. In addition we develop this module by that the users can provide the Ratings.

Rating Prediction

• In this module, we develop the option of providing the Rating by the Social User. In this Rating Prediction a user can rating the items it shows in star based model. The interactions of group memberships determine if a user will connect with another user (i.e.,link prediction) or be interested in a target item. However, theempirical results show that this model is better at link prediction than rating prediction.
• The most popular and widely studied recommendation models are matrix factorization based models which aim to factorize the user item rating matrix into two low-rank user-feature and item feature matrices. Then the predictions can be generated by the inner products of user- and item-specific latent feature vectors.
• Although a user’s rating to a certain item is mainly determined by the intrinsic attributes (or properties, features) of the item in question and how she appreciates these features, some extrinsic attributes may also have a non-negligible influence on the user’s ratings. In this work, we focus on the influence of social trust in rating prediction, i.e., the influence of trust neighbors on an active user’s rating for a specific item, a.k.a. social influence.

Item Recommendation

• In this module, we develop the Item Recommendation. Generally, in social rating networks a user can label (add) other usersas trusted friends and thus form a social network. Trust isnot symmetric; for example, users u1 trusts u3 but u3 doesnot specify user u1 as trustworthy. Besides, users can rate aset of items using a number of rating values, e.g., integersfrom 1 to 5. These items could be products, movies, music,etc. of interest.
• The recommendation problem in this workis to predict the rating that a user will give to an unknownitem, for example, the value that user u3 will give to itemi3, based on both a user-item rating matrix and a usertrust matrix. Other well-recognized recommendationproblems include for example top-N item recommendation.

A Trust-Based Recommendation Model

• In this module first mathematically define the recommendation problem in social rating networks, and then introduce the TrustSVD model.
• In the cold-start situations where users may have only rated a few items, the decomposition of trust matrix can help to learn more reliable user-specific latent feature vectors than ratings-only matrix factorization. In the extreme case where there are no ratings at all for some users, ensures that the user-specific vector can be trained and learned from the trust matrix. In this regard, incorporating trust in a matrix factorization model can alleviate the cold start problem. By considering both explicit and implicit influence of trust rather than either one, our model can better utilize trust to further mitigate the data sparsity and cold start issues.

SYSTEM REQUIREMENTS:

HARDWARE REQUIREMENTS:

• System : Pentium Dual Core.
• Hard Disk : 120 GB.
• Monitor : 15^“ LED
• Input Devices : Keyboard, Mouse.
• Ram : 1 GB

SOFTWARE REQUIREMENTS:

• Operating system : Windows 7.
• Coding Language : JAVA/J2EE
• Tool : Netbeans 7.2.1
• Database : MYSQL