bridging recommendation and adaptation gaf-twittomender compliance case-study john hannon, evgeny...
TRANSCRIPT
Bridging Recommendation Bridging Recommendation and Adaptationand AdaptationGAF-Twittomender GAF-Twittomender compliance case-studycompliance case-study
John Hannon, Evgeny Knutov, Paul De John Hannon, Evgeny Knutov, Paul De Bra, Bra, Mykola Pechenizkiy, Barry Smyth, Mykola Pechenizkiy, Barry Smyth, Kevin McCarthyKevin McCarthy
Agenda
• AH systems evolution, GAF• AH process modelling and GAF building
blocks• GAF-RecSys compliance• Twittomender highlights• Twittomender case-study• Lessons learned• Further Twittomender extension
/ Department of Computer Science PAGE 218-04-23
AH Systems and GAF
/ Department of Computer Science PAGE 318-04-23
• Generalize AHS functionality in GAF reference model
(incl. systems which are not labelled as AH, e.g. RecSys)• Enhance GAF layered structure with the process• Enrich GAF with new approaches, techniques
Adaptation process modeling
/ Department of Computer Science PAGE 418-04-23
Classic loop user modeling - adaptation
Munich Model: Lifecycle of Adaptation
Generic Adaptivity Model
GOMAWE
Classification of AH methods and techniques; adaptation process
/ Department of Computer Science PAGE 518-04-23
• Classification of AH methods and techniques integrated with adaptation process
• Basis for the AHS layered structure
GAF layered model
/ Department of Computer Science PAGE 618-04-23
• GAF aligns the order of the layers in the system according to the classification of AH methods and techniques
• Rotate layered structure of GAF and match with adaptation process flowcharts
GAF layered structure building blocks
GAF ‘sequence chart’ – RecSys compliance
/ Department of Computer Science PAGE 718-04-23
Having trouble finding people to follow on Twitter?
/ Department of Computer Science PAGE 818-04-23
• Syncing a users account and producing followee recommendations through a range of collaborative (3) and content-based (4) strategies
Twittomender architecture
/ Department of Computer Science PAGE 918-04-23
Client-Server architecture
Build on top of Twitter API and Lucene (full features text search engine)
GAF – Twittomender compliance
/ Department of Computer Science PAGE 1018-04-23
GAF – Twittomender compliance (cont.)
• Resource Model – Tweets (via Twitter API)• Goal Model – Twitter Profile login / Search query• User Model – Twittomender User profiler• Domain Model – Lucene index of users’ tweets• Group Model – Collaborative twitter profile
(followers) • Application Model – Twittomender framework• Adaptation Model – Retrieval module (Lucene)• No Twittomender Contextual info
/ Department of Computer Science PAGE 1118-04-23
Lessons Learned
• Helped to identify Twittomender missing parts / extension possibilities• Recommendation of resources of a different
nature
• (elaborate) User Modelling
• Contextualization
• Further Twittomender Extensions (UM interoperability)
/ Department of Computer Science PAGE 1218-04-23
UM interoperability in Twittomender
/ Department of Computer Science PAGE 1318-04-23
/ Department of Computer Science PAGE 1418-04-23
(to be continued)
Thanks!and
Questions?