A contextual bayesian user experience model for scholarly recommender systems / Zohreh Dehghani Champiri
Scholarly recommender systems attempt to narrow down the number of research resources and predict availability of unknown resources to assist scholars with their scholarly tasks. Studies point out that the embedding of the recommending methods in the user experience dramatically affects the value to...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2019
|
| Subjects: | |
| Online Access: | http://studentsrepo.um.edu.my/12479/2/Zohreh_Deghani.pdf http://studentsrepo.um.edu.my/12479/1/Zohreh_Dehghani.pdf http://studentsrepo.um.edu.my/12479/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Scholarly recommender systems attempt to narrow down the number of research resources and predict availability of unknown resources to assist scholars with their scholarly tasks. Studies point out that the embedding of the recommending methods in the user experience dramatically affects the value to the users. Besides, researchers state that factors such as personal and situational characteristics, mostly considered as contextual data, affect the user experience of recommender systems. They started to improve classical recommending methods by modelling contextual data. It has been emphasised, contextual modelling plays a crucial role in recommendations because it can present the status of people, places, objects and devices in the environment. Hence, incorporating contextual data is an effective approach to enhance personalisation, which results in higher efficiency levels of user experience. However, it is not easy to decide which contextual data must be incorporated into scholarly recommender systems. The irrelevant contextual data might have a negative impact and lead to false reasoning models and irrelevant recommendations. Consequently, users lose their trust and stop using the system. Moreover, using too much contextual data leads to computational complexity and ambiguity in the system. Therefore, it requires formulating informed estimations about the influence of certain contexts before exploiting the naturalistic environments. This research aims to first investigate how contexts influence users’ experience of scholarly recommenders and predict the relevant contexts, and then exploits the predicted contexts
to develop a Bayesian user model which can be embedded in the recommending process. Additionally, a user interface for recommendation presentation is designed. Finally, the proposed user model and user interface are evaluated. The empirical results showed that there is strong relation between the user’s contexts and paper quality and user interface design adequacy as well as user interaction design adequacy. The empirical results have been exploited to develop a suitable Bayesian user model and user interface for scholarly recommender systems.
|
|---|