| Online social networks offer researchers almost total access to large amounts of data. This data can be utilized to analyze user profile and identify how similar subsets of users are. In this thesis, degree of closeness/ interaction level can be identified by ranking the users based on similarity score. This similarity is measured on the basis of social, geographic, educational, professional, shared interests, pages liked, mutual interested groups or communities and mutual friends. Of all the techniques used in predicting the similarity between profiles, we use the string similarity metrics and find a new similarity score thus setting a threshold which is efficient and simple in implementation though the analysis is done using large network of Facebook users.;The proposed technique addresses the problem of matching user profiles in its globality by providing a suitable matching framework able to consider all profile's attributes and finding the similarity by new ways of string metrics. The proposed work will be able to discover the biggest possible number of profiles that are similar to the target user profile, which the existing techniques are unable to detect. In our work, attributes were assigned weights manually; string and semantic similarity metrics were used to compare attributes values thus predicting the most similar profiles. Profile based similarity show the exact relationship between users and this similarity between user profiles reflects closeness and interaction between users. |
