Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Towards aligning multi concern models via nlp

323 views

Published on

The design of large-scale complex systems
requires their analysis from multiple perspectives, often
through the use of requirements models. Diversely
located experts with different backgrounds (e.g., safety,
security, performance) create such models using differ-
ent requirements modeling languages. One open chal-
lenge is how to align these models such that they cover
the same parts of the domain. We propose a technique
based on natural language processing (NLP) that ana-
lyzes several models included in a project and provides
suggestions to modelers based on what is represented
in the models that analyze other concerns. Unlike
techniques based on meta-model alignment, ours is
flexible and language agnostic. We report the results
of a focus group session in which experts from the air
traffic management domain discussed our approach.

Published in: Software
  • Be the first to comment

  • Be the first to like this

Towards aligning multi concern models via nlp

  1. 1. TOWARDS ALIGNING MULTI- CONCERN MODELS VIA NLP MODRE’17 F. Başak Aydemir Fabiano Dalpiaz
  2. 2. COLLABORATIVE MODEL-DRIVEN DEVELOPMENT • MULTIPLE EXPERTS/MODELERS • DIVERSE LOCATIONS • DIFFERENT TIME-ZONES • CONCERN SPECIFIC JARGON • THE SAME DOMAIN Towards Aligning Multi-Concern Models via NLP 2
  3. 3. CASE STUDY: AIR TRAFFIC MANAGEMENT CONCERN LANGUAGE LOCATION DOMAIN ONTOLOGY Safety Fault Trees United Kingdom AIRM* Organization Goal Models Netherlands AIRM … … … … Towards Aligning Multi-Concern Models via NLP 3 *Air Traffic Management Information Model (www.airm.aero) Example set up of a European air traffic management project: • PACAS is a tool-supported process that fosters the active collaboration among heterogeneous stakeholders • Unlike traditional, informal enterprise architectures PACAS relies on gamification and automated reasoning to formally align the multiple views
  4. 4. SOLUTION ARCHITECTURE • WEB-BASED • ASYNCHRONOUS • COLLABORATIVE • NLP-BASED • (MODELING) LANGUAGE-AGNOSTIC • DOMAIN ONTOLOGY Towards Aligning Multi-Concern Models via NLP 4
  5. 5. INTERACTION BETWEEN MODELERS Towards Aligning Multi-Concern Models via NLP 5
  6. 6. COMMIT A MODEL Towards Aligning Multi-Concern Models via NLP 6
  7. 7. ANALYZING COMMITS 1. ELEMENT LABELS {AIR TRAFFIC CONTROLLER SUPERVISED} {SECTORS SPECIFIED} 2. NOUNS FROM THE LABELS {AIR, TRAFFIC, CONTROLLER, SECTOR} 3. KEEP TRACK OF NOUN FREQUENCIES Towards Aligning Multi-Concern Models via NLP 7
  8. 8. INTERACTION BETWEEN MODELERS Towards Aligning Multi-Concern Models via NLP 8
  9. 9. SUGGESTING CONCEPTS 1. MISSING NOUN {SECTOR} 2. RELATED CONCEPTS FROM DOMAIN ONTOLOGY {MINIMIM SECTOR ALTITUDE, SECTOR CONFIGURATION PLAN, CONTROL SECTOR, ….} 3. FILTER RESULTS {MINIMUM SECTOR ALTITUDE, SECTOR CONFIGURATION PLAN} Towards Aligning Multi-Concern Models via NLP 9
  10. 10. INTERACTION BETWEEN MODELERS Towards Aligning Multi-Concern Models via NLP 10
  11. 11. EVALUATION • A FOCUS GROUP DISCUSSION WITH ATM EXPERTS • POSITIVE FEEDBACK FOR THE NLP-BASED SUPPORT • DIVERSE PREFERENCES (NUMBER OF SUGGESTIONS, HIGH-LEVEL VS DETAILED SUGGESTIONS, TIMING OF THE SUGGESTIONS) • HEURISTICS NEEDED • FINE TUNING NEEDED • THE DIRECTION OF THE FEEDBACK SHOULD BE CONSIDERED IN SOME SETTINGS (E.G., SECURITY MODEL FEEDS THEM ALL) Towards Aligning Multi-Concern Models via NLP 11
  12. 12. QUESTIONS TO THE AUDIENCE • ANY INTERESTING CASES ABOUT SOFTWARE DEVELOPMENT WHERE THE APPROACH CAN BE USED? • WHAT RISKS MAY THE APPROACH INTRODUCE FOR THE MODELING PROCESS? Towards Aligning Multi-Concern Models via NLP 12
  13. 13. CONTACT US Towards Aligning Multi-Concern Models via NLP 13 F. Başak Aydemir Fabiano Dalpiaz f.b.aydemir@uu.nl f.dalpiaz@uu.nl @aydemirfb @FabianoDalpiaz This work has received funding from the SESAR Joint Undertaking grant agreement No 699306 under European Union’s Horizon 2020 research and innovation programme. www.pacasproject.eu @pacasproject
  14. 14. IMPORTANT DATES • ABSTRACTS SEPTEMBER 25, 2017 • PAPERS OCTOBER 2, 2017 • CONFERENCE MARCH 19-22, 2018 www.refsq.org/2018

×