Broadway : a recommendation computation approach

based on user behaviour similarity

Brigitte Trousse, Rushed Kanawati

Acion AID, INRIA Sophia Antipolis
e-mail : Brigitte.Trousse@.inria.fr, Rushed.Kanawati@inria.fr


Recommendation systems have gained a lot of attention recently especially in the field of on-line information retrieval (IR) systems (e.g. searching documents on the web). Recommenders are traditionally partitioned into two main families: content-based recommenders and collaborative filtering ones. Systems of the first type recommend items or actions to the user depending on an evaluation of the user own past actions, while those belonging to the second family recommend to a user items positively evaluated by other similar users. In this work we describe a new recommendation approach, called the Broadway approach, where the system recommends to a user what have satisfied other users (or eventually similar users) that have behaved similarly to that user.

Following the Broadway approach, the user interactions with the application are saved in a log-like file. This log file contains a set of time series, each holds the evolution with time of a variable that is said to be relevant to describe the user behavior. Obviously the choice of these variables depends on the application field. Time series are grouped into records that correspond to a user session with the system within a well specified period of time which has a well defined semantic in the application. Case-based Reasoning (CBR) technology is used to implement the Broadway approach. CBR is a problem solving technology where, in order to find the solution to a current problem, one looks for a similar problem in an experience base, takes the solution from the past and use it for a starting point to find a solution to the current problem. A case is a contextualized piece of knowledge representing an experience. It is generally composed of two main parts : the problem and the solution.

Also, from records, we extract potential useful experiences, called potential cases, by using a case template issued from domain experts. Such a template identifies useful situations from saved sessions (or records) as well as the solutions (i.e. recommendations) suggested by these situations. The situation, formed of a sub-history of the user session, is used to retrieve past situations that match the most the current user behaviour and could explain the current behaviour in that session. The definition of a situation (i.e. indices that are said to be relevant to explain the user behaviour) depends on the nature of recommendations we want to provide. Figure  1 illustrates the classical CRB system cycle applied by the Broadway approach.
 
 

Figure 1. The Broadway recommendation computation approach cycle

The Broadway approach is already used to implement two concrete applications that aim at facilitating the task of information searching on the Web. The first application is a Web browsing advisor, called Broadway V1,  while the second, called Broadway-QR,  is a query refinement recommender integrated in a query based web searching system.