WalkonPaths offers reliable recommendations to user by handling the information overload on online communities, and protects him from spamming, malicious users and misreporting
WalkonPaths offers reliable recommendations to users by handling the information overload on online communities, and protects them from spamming, malicious users and misreporting. This is being achieved by using in house implemented algorithms for Internet data crawling, Internet data analysis, user behavior analysis and algorithms for recommendations/predictions.
Trust, similarity and reputation are three basic ingredients for more accurate recommendations on online communities. Trust of a user indicates his trustworthiness, similarity between two users indicates their proximity, and reputation of a user indicates the opinion of the other users about that user. By using these three components, we provide recommendations on products that user might is interested in, e.g. travel destinations that fit with user profile, users that can provide to other users mentorship on specific subjects, etc. Trust, similarity and reputation answer on the following questions: what is the validity of a user that provides a review? Is this user similar with you? How well is he reputed on the online community?
The basic functionality of WalkonPaths is presented in the following image. A user asks for recommendations through a query, which passes to the recommender system. Using past user interactions, the recommender system computes the three basic components: their trust, similarity and reputation. Using those components, WalkonPaths returns to the end user a set with the computed recommendations.
WalkonPaths is based on the community structure. A community is represented as a graph where users and items are the nodes and their relationships are the edges. The edges can be weighted and have several features e.g. time of creation, number of interactions with other nodes, etc.
The competitive advantages of WalkonPaths principle are that it: