HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008 Interface Model Elicitation from  Textual Scenarios  Christophe Lemaigre, Josefina Guerrero, Jean Vanderdonckt Université catholique de Louvain (UCL) Louvain School of Management (LSM) Information Systems Unit (ISYS) Place des Doyens, 1 – B-1348 Louvain-la-Neuve (Belgium) http://www.isys.ucl.ac.be/
Introduction and motivations Model  Elicitation Consists of The identification of model elements From Textual scenario First step of a model-driven engineering process Selection of several models : user, task, domain, organization, resource and job Characterizing  the concepts used in the development life cycle of user interfaces for Worfklow Information Systems HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
The underlying ontology Reduced view Task : piece of work (same resource, location, time period) Organizational unit : physical location, equipped with resources  User stereotype : human being HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008 Organizational Unit Job Task 1..* 1..* 1..* 1..* Task Resource User Stereotype Material Immaterial Process Workflow 1..* 1..* * 0..1 0..1 * 0..1 * 1..* 1..* 1..* 1..* isOrganizedInto ► isOrderedIn ► 1..* 1..* Object Method Manipulates ► Invokes ► * * * *
The underlying ontology Expanded view HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
Related  work Some other tools use model elicitation at some level  U-Tel, ConcurTaskTress, T2T, Garland et al. Brasser & vanderLinden Shortcomings : Focused on a single model No attribute elicitation Result that can hardly be exploited HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
Methodology and tool support We developped an elicitation methodology based on three levels Manual classification Dictionary-based classification Semantic understanding And implemented the first and second one in a tool, made of A text edition part, with syntactic coloration Trees in which model elements are dispatched HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
Tool support Model Elicitation Tool HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
Level 1: manual classification Definition : Program user does the elicitation job Without the help of an automated process Method : Selection of a piece of text from the scenario Choose the appropriate model and object type HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
Level 1: manual classification Tool : elicitation of a task HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008 1 2 3
Level 1: manual classification Advantages :  Accurate result Easier to implement than automated elicitation No need of classification datas Inconvenients :  Fastidious for the user Time costly HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
Level 2: dictionary-based classification Definition : Underlies on a set of predefined terms that will be automatically extracted and identified as model objects Two kinds of dictionaries : Generic dictionary, which is domain-independant  Specific dictionary, linked with a definite domain Method : Improved pattern-matching process :  Based on the recognition of phrases That are associated with their model definition Plural forms and conjugation are taken into account (e.g. to provide // providing) HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
Level 2: dictionary-based classification Tool : elicitation of jobs using a dictionary HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
Level 2: dictionary-based classification Advantages :  Processing speed No human intervention needed Inconvenients :  Lack of precision, some elements being poorely classified due to the fact it is context-independant No relations between elements (e.g. hierarchy) HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
Level 3: toward semantic understanding Definition : Try to approximate natural language understanding Method : Using syntactic tagging, semantic tagging and chunk parsing. Detection of Concepts such as task types or attribute types Relationships between model elements No tool support currently HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
Level 3: toward semantic understanding Concrete example “ An accountant receives taxes complaints, but he is also in charge of receipts perception” Model elements : Task : receive taxes complaints Task : charge of receipts perception Job : accountant Relation “performed by” between those tasks an the job  Temporal operator : concurrency for the tasks, used by default HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
Level 3: toward semantic understanding Advantages :  Expressivity, being able to deduce relationships between model elements Automatic treatement Inconvenients :  Difficult to implement Natural language understanding is a field of informatics research that needs a lot of work and improvement HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
After model elicitation Once elicitation job is done, some treatments can be performed Use of syntactical coloration allowing the user to check its work Verification of the compliance with some desirable quality properties UsiXML export, allowing to use tools like IdealXML or FlowiXML to edit models and derivate user interfaces HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
Conclusion and future work Methodology and tool support Combination of three complementary methods Allowing elicitation of elements from several models and relations between those elements Oriented towards user-interfaces generation for workflow information systems Implemented in a tool, using Usi-XML standard to export its result Future works : Advanced visualisation (e.g carrousel) Take into account inter-model relationships Refine the third level towards a more natural language understanding HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
Thank you very much for your attention HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008 For more information and downloading, http://www.isys.ucl.ac.be/bchi   http://www.usixml.org User Interface eXtensible Markup Language http://www.similar.cc European network on Multimodal UIs Special thanks to all members of the team!

Interface Model Elicitation from Textual Scenarios

  • 1.
    HCIS’2008 – Milan,September 8-9th, 2008, during IFIP World Congress 2008 Interface Model Elicitation from Textual Scenarios Christophe Lemaigre, Josefina Guerrero, Jean Vanderdonckt Université catholique de Louvain (UCL) Louvain School of Management (LSM) Information Systems Unit (ISYS) Place des Doyens, 1 – B-1348 Louvain-la-Neuve (Belgium) http://www.isys.ucl.ac.be/
  • 2.
    Introduction and motivationsModel Elicitation Consists of The identification of model elements From Textual scenario First step of a model-driven engineering process Selection of several models : user, task, domain, organization, resource and job Characterizing the concepts used in the development life cycle of user interfaces for Worfklow Information Systems HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 3.
    The underlying ontologyReduced view Task : piece of work (same resource, location, time period) Organizational unit : physical location, equipped with resources User stereotype : human being HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008 Organizational Unit Job Task 1..* 1..* 1..* 1..* Task Resource User Stereotype Material Immaterial Process Workflow 1..* 1..* * 0..1 0..1 * 0..1 * 1..* 1..* 1..* 1..* isOrganizedInto ► isOrderedIn ► 1..* 1..* Object Method Manipulates ► Invokes ► * * * *
  • 4.
    The underlying ontologyExpanded view HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 5.
    Related workSome other tools use model elicitation at some level U-Tel, ConcurTaskTress, T2T, Garland et al. Brasser & vanderLinden Shortcomings : Focused on a single model No attribute elicitation Result that can hardly be exploited HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 6.
    Methodology and toolsupport We developped an elicitation methodology based on three levels Manual classification Dictionary-based classification Semantic understanding And implemented the first and second one in a tool, made of A text edition part, with syntactic coloration Trees in which model elements are dispatched HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 7.
    Tool support ModelElicitation Tool HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 8.
    Level 1: manualclassification Definition : Program user does the elicitation job Without the help of an automated process Method : Selection of a piece of text from the scenario Choose the appropriate model and object type HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 9.
    Level 1: manualclassification Tool : elicitation of a task HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008 1 2 3
  • 10.
    Level 1: manualclassification Advantages : Accurate result Easier to implement than automated elicitation No need of classification datas Inconvenients : Fastidious for the user Time costly HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 11.
    Level 2: dictionary-basedclassification Definition : Underlies on a set of predefined terms that will be automatically extracted and identified as model objects Two kinds of dictionaries : Generic dictionary, which is domain-independant Specific dictionary, linked with a definite domain Method : Improved pattern-matching process : Based on the recognition of phrases That are associated with their model definition Plural forms and conjugation are taken into account (e.g. to provide // providing) HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 12.
    Level 2: dictionary-basedclassification Tool : elicitation of jobs using a dictionary HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 13.
    Level 2: dictionary-basedclassification Advantages : Processing speed No human intervention needed Inconvenients : Lack of precision, some elements being poorely classified due to the fact it is context-independant No relations between elements (e.g. hierarchy) HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 14.
    Level 3: towardsemantic understanding Definition : Try to approximate natural language understanding Method : Using syntactic tagging, semantic tagging and chunk parsing. Detection of Concepts such as task types or attribute types Relationships between model elements No tool support currently HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 15.
    Level 3: towardsemantic understanding Concrete example “ An accountant receives taxes complaints, but he is also in charge of receipts perception” Model elements : Task : receive taxes complaints Task : charge of receipts perception Job : accountant Relation “performed by” between those tasks an the job Temporal operator : concurrency for the tasks, used by default HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 16.
    Level 3: towardsemantic understanding Advantages : Expressivity, being able to deduce relationships between model elements Automatic treatement Inconvenients : Difficult to implement Natural language understanding is a field of informatics research that needs a lot of work and improvement HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 17.
    After model elicitationOnce elicitation job is done, some treatments can be performed Use of syntactical coloration allowing the user to check its work Verification of the compliance with some desirable quality properties UsiXML export, allowing to use tools like IdealXML or FlowiXML to edit models and derivate user interfaces HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 18.
    Conclusion and futurework Methodology and tool support Combination of three complementary methods Allowing elicitation of elements from several models and relations between those elements Oriented towards user-interfaces generation for workflow information systems Implemented in a tool, using Usi-XML standard to export its result Future works : Advanced visualisation (e.g carrousel) Take into account inter-model relationships Refine the third level towards a more natural language understanding HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008
  • 19.
    Thank you verymuch for your attention HCIS’2008 – Milan, September 8-9th, 2008, during IFIP World Congress 2008 For more information and downloading, http://www.isys.ucl.ac.be/bchi http://www.usixml.org User Interface eXtensible Markup Language http://www.similar.cc European network on Multimodal UIs Special thanks to all members of the team!