Enhanced content-based filtering algorithm using artificial bee colony technique

Mahmoud, Dima Saber (2014) Enhanced content-based filtering algorithm using artificial bee colony technique. [Dissertation (University of Nottingham only)]

[img] PDF - Registered users only - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
Download (2MB)

Abstract

Recommender systems can guide the users in a tailored way to interesting objects in a large space of possible options. The Content-based Filtering (CBF) approach is one of the most widely adapted to date. It basically analyses a set of textual descriptions of items. These items are already evaluated by an interactive user in prior steps. Then it builds a model or profile of this user

accordingly. The profile is then exploited to suggest a new item. Unfortunately, filtering module in the Content-based filtering is mainly based on one item at a time rather than exploring a complete set of available items. Our concern here is; how this component can recommend list of items to a user from huge amount of data. The main objective here is to enhance the Content-based Filtering algorithm in order to explore a huge data set and get a list of recommendations rather than just rating an item. For this, an Artificial Bee Colony technique has been adapted and applied to the Content-based Filtering method. ABC is one of the efficient evolutionary Computing techniques that are used in solving optimization problems.

One of the major evolutionary techniques is Artificial Bee Colony (ABC) which is inspired from bee life and how it behaves in group. It is one of the efficient techniques to be used in optimization problems. In our research to enhance the Content-based filtering algorithm, we applied ABC technique in order to get a more advanced algorithm that can exploit the user interests and access large amount of data to return more relevant recommendations for the user.

We designed and implemented a case study for proving the results of the enhanced algorithm. With comparing results of the enhanced algorithm with those of the original one, we got that the proposed algorithm has showed better

results. The new enhanced algorithm has the capability of suggest relevant recommendations to the user with average rate 6.5 in case of high detailed user profile and 5.6 in case of medium detailed profile.

Item Type: Dissertation (University of Nottingham only)
Depositing User: Gonzalez-Orbegoso, Mrs Carolina
Date Deposited: 13 Nov 2015 10:13
Last Modified: 19 Oct 2017 15:06
URI: https://eprints.nottingham.ac.uk/id/eprint/30759

Actions (Archive Staff Only)

Edit View Edit View