Shopster – A Social Mobile App

This article presents 3 weaknesses for Collaborative Filterings (CF) algorithms. These are: data sparsity (difficult to calculate user similarity because of very few commonly rated items), cold-start users (new users with no ratings – or no friends in our case) and malicious users attacking Recommender Systems (e.g. by copy-profile attack). They conduct experiments using a large (real) database from a website and prove that having a trust metric between users is much more accurate than common CF algorithms. We do not necessarily have to include a trust value for every relationship in Shopster but we can incorporate trust as a measure of how trustworthy a user is, based on their divergence from actual average ratings but also on how helpful their posts and comments (if any) are rated by other users.

Source: Trust-aware Recommender Systems

This is an article presenting a website, FilmTrust, for rating and recommending movies to the users. The most important idea that we can get for Shopster out of this article is the way they propose for calculating user trust in the system. Each user is able to assign a trust value to their friends (e.g. Alice trusts Bob 9 out of 10) and therefore their recommended rating for a movie can be predicted by observing their most trusted friends’ rating, using TidalTrust.

For example if Alice trusts Bob 9 and Mary 3, and if Bob rated a movie with 4 stars while Mary rated the same movie with 2 stars, the recommended rating for Alice would be: [(9 x 3) + (3 x 2)] / (9 + 3) = 3.5. This method was also compared to a recommended rating using Pearson Correlation coefficient (Automatic Collaborative Filtering) for calculating their nearest neighbour’s prediction as well as a simple total average for a movie, and found to be more accurate than both of them.

We could use this method of user trust between users in Shopster with the only drawback being that this method is not efficient if a user does not have any friends.

Article source: FilmTrust: Movie Recommendations using Trust in Web-based Social Networks

It is a paper that was published in 2002 in the First international conference on mobile business by Bernhard Kolmel and Spyros Alexakis. It analyses how a variety of services can exploit knowledge about the location of a user of a mobile device by targeting advertisements to specific users that are more likely to be affected. Also it explains how various technologies such as GPS, 3G, etc can be used in location based advertisement. In my opinion this is an important article which must be read carefully by people that are interested in this area, in order to understand some basic aspects of location based advertising. It can be used by our application in order to target advertisements based on the location of the application users.

Source: Location Based Advertisement

This article is about a social-based mobile application designed to analyse the user’s messaging activity to identify the user’s closest social contacts.

The main purpose of that mobile application maybe is not too similar to our project but it examines social concepts that we should consider as well. For instance,  people make judgments about whom to share information with based more on the identity of the recipient of the information than on the situation within the information was sought. The users can be associated using their given preferences thus creating their own social group. Therefore, it would be great to contain such rules in our application improving the social interaction of all involved users.

Additionally, Friendlee’s  users can easily share key aspects of her context such as their location at different granularities (country, city or GPS-based street address) and status (e.g available). Such ambient awareness of people’s closest connections helps them feel emotionally close and also facilitates communication. To protect privacy, people have access to this context-sharing functionality only for a selected group of users.

While browsing a friend’s connections, people also see their preferred businesses, getting implicit recommendations about e.g. the dentist their friend likes to go to. In addition, Friendlee allows users to search their social network for people and businesses. Search results are ranked by social distance from the user. Obviously, that feature can be used in the terms of our mobile’s functionality. For example, recommendations will be made according to similar preferences of users that have already interested in a same category of products.

Reference: Ankolekar, A., Szabo, G., Luon, Y., Huberman, B.A., Wilkinson, D., Wu, F.: Friendlee: A Mobile Application for Your Social Life. In: 11th International Conference on Human-Computer Interaction with Mobile Devices and Services, Article No. 27 (2009)

Available: http://www.hpl.hp.com/research/scl/papers/friendlee/friendlee-old.pdf