On Friday the fourth of July we presented the outcomes of our SIRUP project at the Network Institute end of year presentation. Central to our presentation were the results of the testing of the theoretical model and the three preliminary studies (complexity, familiarity, and conflict). If you have missed our presentation or would like to take another look at it, you can check out our presentation here: Network Institute SIRUP end of year presentation.
For the SIRUP project we conducted a total of three preliminary studies in order to try to identify successful indicators for complexity, familiarity, and conflict. In addition to the measures for these three factors, we included measures to enable us to test the theoretical model we developed based on the work of Daniel E. Berlyne and David Sylvia (see Figure 1).
Figure 1. SIRUP theoretical model
In the third preliminary study of SIRUP we tried to identify indicators that predict the perceived conflict of a television program. In the current context, conflict refers to incongruency in peoples’ evaluation of a program. During this project we work with broadcasting data from BBC’s ViSTA-TV project. Based on the available data from this project and external public sources, we identified two possible indicators for conflict (see Figure 1):
- Variance (SD) in IMDB user ratings
- Variance (SD) in BBC user ratings
Figure 1. Testing the selected indicators for conflict
In the second preliminary study of SIRUP we tried to identify indicators that predict the perceived familiarity of a television program. During this project we work with broadcasting data from BBC’s ViSTA-TV project. Based on the available data from this project and external public sources, we identified five possible indicators for familiarity (see Figure 1):
- Number of BBC viewers
- Number of Google results
- Number of Facebook likes
- Number of Twitter references
- IMDB user ratings
Figure 1. Testing the selected indicators for familiarity
In the first preliminary study of SIRUP we tried to identify indicators that predict the perceived complexity of a television program. During this project we work with broadcasting data from BBC’s ViSTA-TV project. Based on the available data from this project and the work of Daniel E. Berlyne, we identified three possible indicators for complexity (see Figure 1):
- Number of credits (people involved in the production of a program)
- Number of actors
- Number of categories (amount of formats and genres of a program)
Figure 1. Testing the selected indicators for complexity
Today, we will present our SIRUP project at the Network Institute poster session. Central to our presentation is the question of how we will create the experience of serendipity in a user through a recommendation provided by the recommender system. If you want to find out, come check out our poster (VU University Amsterdam, April 15, 16.00-17.30, Intertain Lab, W&N Building S1.11) or take a look at our poster below.
Serendipity is making a pleasant and relevant discovery that was unexpected. In SIRUP we argue that a serendipitous recommendation primarily induces interest in a user. Hence, serendipitous recommendations are those that trigger interest in users.
Building on the classic theory of interest by Daniel E. Berlyne and the follow-up work by Paul Silvia, we argue that user interest, and thus serendipity, is determined by two things: novelty (inherent to the recommendation) and coping potential (inherent to the user). Novelty, in turn, is determined by the complexity, familiarity and conflict characteristics of the recommended item.
Based on this logic, we present SIRUP’s theoretical model:
How can we enhance serendipity in user recommendations? After an intense period of theorizing, conceptualizing, and preparing the stimulus material, the first preliminary study of the SIRUP project is online now! The first data begin to trickle in and soon we will be able to present the very first results of our project. Want to know more about the things that we have been working on? Come and take a look (and oh yes: participate) here.
Creating serendipity (i.e. “pleasant surprises for users”) is a primary goal of intelligent recommender systems. This project proposes an interdisciplinary approach to enhance the serendipity of TV recommendations that combines complementary knowledge from three disciplines – Computer Science, Language & Cognition and Communication Science.
The project examines the “back-end” or algorithms behind serendipitous TV recommendations (Computer Science), the “front-end” or the actual display of these recommendations (Language & Cognition), and the “effect” on users’ perceptions and satisfaction (Communication Science).