User profile retrieval in social networks based on facial recognition and personal information

In the last decade social network sites have been rapidly growing. The sharing of photos has become an important part of these sites. The amount of information willingly shared by people on their profiles there is astonishing. In this paper we show that personal context information and a photo is a powerful combination to retrieve a person. Using our method it is possible for users to identify someone with only a photo and some personal information out of a set of hundreds of persons. The created prototype consists of several text fields for input of personal information. The user can select the most similar person from a ranking based on this context info and the similarity of the photo. In a user-evaluation we compare our system to the use of only facial recognition and a photo without knowledge of personal context information. This showed that using known contextual information from outside a photo on top of face recognition can increase the retrieval rate.