Personalized Recommendation Combining User Interest and Social Circle

ABSTRACT:

With the advent and popularity of social network, more and more users like to share their experiences, such as ratings,reviews, and blogs. The new factors of social network like interpersonal influence and interest based on circles of friends bringopportunities and challenges for recommender system (RS) to solve the cold start and sparsity problem of datasets. Some of thesocial factors have been used in RS, but have not been fully considered. In this paper, three social factors, personal interest,interpersonal interest similarity, and interpersonal influence, fuse into a unified personalized recommendation model based onprobabilistic matrix factorization. The factor of personal interest can make the RS recommend items to meet users’ individualities,especially for experienced users. Moreover, for cold start users, the interpersonal interest similarity and interpersonal influence canenhance the intrinsic link among features in the latent space. We conduct a series of experiments on three rating datasets: Yelp,MovieLens, and Douban Movie. Experimental results show the proposed approach outperforms the existing RS approaches.

PROJECT OUTPUT VIDEO:

EXISTING SYSTEM:

The quality of recommendations and usability of sixonline recommender systems was examined. Theresults show that the user’s friends consistently providedbetter recommendations. For example 90% of users believethe book recommended is good from friends, 75% of usersbelieve that the recommendation is useful from friends.This research shows that the interpersonal influence isimportant in social media. Java et al.had analyzed alarge social network in a new form of social media knownas micro-blogging. It has a high degree correlation andreciprocity, indicating close mutual acquaintances amongusers. They had identified different types of user intentionsand studied the community structures. Categorizing friendsinto groups (e.g. family, co-workers) would greatly benefitthe adoption of micro-blogging platforms to analyze userintentions. That is to say user intentions or interests can bereflected by those of its friends.

DISADVANTAGES OF EXISTING SYSTEM:

• Do all users actually need the relation-ship on the social networks to recommend items? Does the relationship submerge user’s personality, especially for the experienced users? It is still a great challenge to embody user’s personality in RS, and it is still an open issue that how to make the social factors be effectively integrated in recommendation model to improve the accuracy of RS.

PROPOSED SYSTEM:

In this paper, three social factors, personal interest, inter-personal interest similarity, and interpersonal influence,fuse into a unified personalized recommendation modelbased on probabilistic matrix factorization. The personality is denoted by user-item relevance of user interest to thetopic of item. To embody the effect of user’s personality,we mine the topic of item based on the natural item category tags of rating datasets. Thus, each item is denoted by acategory distribution or topic distribution vector, which canreflect the characteristic of the rating datasets. Moreover, weget user interest based on his/her rating behavior. We thenassign to the effect of user’s personality in our personalizedrecommendation model proportional to their expertise levels. On the other hand, the user-user relationship of socialnetwork contains two factors: interpersonal influence andinterpersonal interest similarity. We apply the inferred trustcircle of Circle-based Recommendation (CircleCon) model to enforce the factor of interpersonal influence. Similarly,for the interpersonal interest similarity, we infer interestcircle to enhance the intrinsic link of user latent feature.

ADVANTAGES OF PROPOSED SYSTEM:

• The factor of user personal interest makes direct connections between user and item latent feature vectors. And the two other social factors make connections between user and his/her friends’ latent feature vectors.
• Personal unique interest is modeled to get an accurate model for the cold start user and user with very few friends and rated items. The impacts of the three factors to the recommendation performances are systematically compared.
• the effect of proposed model to solve the usercold start and sparsity problem

MODULES:

1. User Personal Interest
2. Interpersonal Interest Similarity
3. Recommend User Interested Items

MODULES DESCRIPTION:

1. User Personal Interest:

Besides the trust values between friends in the same category [1], user interest is another significant factor to affectusers’ decision-making process, which has been proved bypsychology and sociology studies [38]. Moreover, Jianget al.[2] demonstrated the effect of ContextMF model with consideration of both individual preference and interpersonalinfluence. However, there are two main differences of theuser interest factor in our model to individual preferencein ContextMF 1) the independence of user interest. Itmeans we can recommend items based on user interest at acertain extent. In other words, we utilize user’s connectionwith the items to train the latent feature vectors, especially for the experienced users. 2) Interest circle inference.Just like CircleCon model [1], we divide the tested socialnetwork into several sub-networks, and each of them corresponds to a signal category of items. Considering the coldstart users who has a few rating records, we use friends’interest in the same category to link user latent featurevector.\

1. User Interest Description:

According to the natural item category tags of ratingdatasets, we can get category distribution of the item, whichcan be seen as the naïve topic distribution of the itemDi.For example, each item has the tags of its category in Yelp.Just like the item The Dakota Barof New York belongs tothe category Night Life, and thenNight Life is one of the category tags of the item. From user’s historical rating datain category c, we summarize all the rated items to measureuser interest.

1. Personal Interest:

Due to the individuality, especially users with many ratingrecords, users usually choose items all by themselves withlittle influence by their friends. However, many previousworks took the circles of friends in social networksto solve the cold start problem. It did work for the cold startusers with a few records, but ignored the individuality forexperienced users. In other words, the relevance of userand item latent feature vector depends on the relevanceof user interestDu and item topicDito a certain extent.More formally, we denote the relevance of useru’s personalinterest to the topic of itemi in our recommendation model.

1. Interest Circle Inference:

Similar to the trust circle inference in CircleCon model, we propose the interest circle inference. The basicidea is that user latent feature vector should be similar tohis/her friends’ latent feature vector based on the similarity of their interest.

2. Interpersonal Interest Similarity:

The personalized recommendation model contains the following three aspects: 1) Interpersonal influence S^c∗_u,v,which means whom you would trust. 2) Interest circle inferenceW^c∗_u,v, which means whose interest is similar to yours.3) User personal interestQ^c∗_u,i, which has effect on whatitems you would interest in.

3. Recommend User Interested Items:

The normalized number of items that user uhas rated inc,which is the factor of a user depends on his/her personalinterest to rate an item. The initial values of U^candP^care sampled from the normal distribution with zero mean.It empirically has little effect on the latent feature matrixlearning. The user and item latent feature vectors U^candP^care updated based on the previous values to insure thefastest decreases of the objective function in each iteration.Note that the step size is a considerable issue. We adjust itto insure the decreases of the objective function in training.

SYSTEM REQUIREMENTS:

HARDWARE REQUIREMENTS:

• System           : Pentium IV 2.4 GHz.
• Hard Disk      : 40 GB.
• Floppy Drive : 44 Mb.
• Monitor          : 15 VGA Colour.
• Mouse             : Logitech
• Ram                 : 512 Mb.

SOFTWARE REQUIREMENTS:

• Operating system : Windows XP/7.
• Coding Language : net, C#.net
• Tool                         : Visual Studio 2010
• Database                : SQL SERVER 2008