11 Towards a Research Agenda for Recommendation Systems in Requirements Engineering
1. Towards a Research Agenda forRecommendation Systems in Requirements Engineering Walid Maalej, TU München <maalejw@cs.tum.edu> AnilKumarThurimella, Harman Becker Automotive Systems
2. 2 Outline Motivation State of the Art Enablers Applications Next Steps 1 2 3 4 5 Towards a Research Agenda for RS in RE
14. 5 Outline Motivation State of the Art Enablers Applications Next Steps 1 2 3 4 5 Towards a Research Agenda for RS in RE
15. Towards a Research Agenda for RS in RE 6 Research Reaped Several Tools* * Happel & Maalej: RSSE08, Potentials And Challenges of Recommendation Systems in Software Development
16. Towards a Research Agenda for RS in RE 7 Limitations of Current Approaches No support for requirements engineering work, no use of requirements as recommendation items 3 Information provision and experience capture not supported Limited pro-activeness, personalization and use of context 2 1 Inflexible, hard-coded knowledge representation and heuristics Centralized and inflexible architecture Limitation in functionality and technology To tape the potentials of RS in RE, a systematic account is required
17. 8 Outline Motivation State of the Art Enablers Applications Next Steps 1 2 3 4 5 Towards a Research Agenda for RS in RE
26. RS using inverse search:Matches the corpus of information providers against a given set of queries Identify information worth sharing Index 3. inSearch 1. Query Queries 4. Share 2. Results * Maalej & Happel: PAKM08, A Lightweight Approach forKnowledgeSharing in Distributed Software Teams
27. 11 Outline Motivation State of the Art Enablers Applications Next Steps 1 2 3 4 5 Towards a Research Agenda for RS in RE
28. Towards a Research Agenda for RS in RE 12 Recommendation Landscape For RE
38. Towards a Research Agenda for RS in RE 18 Challenges of Context Awareness Design How can we model abstract concepts to support unique scenarios? Long-Term Context Can we objectively model a subjective topic like experience? Sessionization How can we “package” context based on the time and content? Six major research challenges Aggregation How can we support different levels of abstractions with different factors? Context Representation Can we efficiently represent data and enable simple and semantic interpretation? Context Factors What should be observed, what is a part of the context?
39. Towards a Research Agenda for RS in RE 19 Open Source Platform: TeamWeaver www.teamweaver.org
40. Towards a Research Agenda for RS in RE 20 Four States of Private and Organizational Knowledge Space There is enough knowledge w.r.t. a certain topic, both locally and in the organization A user has lots of knowledge about a topic, while there is not much knowledge in the organization many Information overload Organizational Information gap few Documents in private space Personal information gap Information shortage few many Documents in organizational space While there is lots of knowledge in the organization, the user does not have much knowledge Few knowledge exists both locally and globally