Aatish - A New Profile-Based Recommendation Services for Mobile Telecom Network Subscribers
Current mobile telecom networks generate massive amounts of subscribers' data consisting of observations on duration of call, time of call, type of call, plans subscribed and other details in addition to network data. Due to the ever increasing degree in complexity of understanding behavioural...
Saved in:
| Published in: | 2015 Second European Network Intelligence Conference pp. 160 - 164 |
|---|---|
| Main Authors: | , , , |
| Format: | Conference Proceeding |
| Language: | English |
| Published: |
IEEE
01-09-2015
|
| Subjects: | |
| Online Access: | Get full text |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Current mobile telecom networks generate massive amounts of subscribers' data consisting of observations on duration of call, time of call, type of call, plans subscribed and other details in addition to network data. Due to the ever increasing degree in complexity of understanding behavioural pattern of subscribers in networks, coupled with specific constraints related to subscribers' preferences, the traditional recommendation approaches are not efficient in meeting customer demands. A new recommendation paradigm is required that uses techniques that exploit user-specific information based on time, location, recent interest, etc., rather than common techniques, to prioritize recommendation. Our work presents the design and specification of Aatish, a profile-based recommendation system as a solution to increase the probability of acceptance of recommendations by individuals. In our approach, recommendation services are modelled using a newly proposed policy-based engine that combines information from subscriber's behavioural aspects of Mobile Telecom Networks (MTN) and Social Media (SM) inputs. Services are also associated with requirements, that is, set of propositions that characterize the execution environment of the service. The architecture of Aatish illustrates policy enhancement that takes into account the context as well as content-aware situations and user preferences that can impact the performances and functionalities of efficient recommendation system. Our study starts by extracting user information from CDRs and social media to generate sequences of recommendations. After observing response from users for the recommendations, appropriate actions are selected, re-ranked and recommended so that the engendered recommendations match recent needs of the users. Moreover, subscriber's ranking preferences have been derived based on the behavioural pattern of users in terms of their mobile phone usage (Recency, Frequency, Monetary) for the benefit of improving the loyalty scores. |
|---|---|
| DOI: | 10.1109/ENIC.2015.32 |