From Kolflow Project

Revision as of 10:40, 22 February 2011 by Khaled (Talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Title : Scenarios and Evaluations

Code : 6

Responsible : GDD

Activities : project evaluation

Start Date : 2011-02-01

End Date : 2014-07-31

Objectives : The objectives of this task are:

  • To define compelling scenarios of man-machine collaboration based on past experiences of partners.
  • To evaluate each year the advance in realization of this scenarios.

Scenarios will be built based on the experience gained by Orpailleur, Silex and Score in the cooking recipes domain through their collaboration in the Taaable project. A second version of the system, based on a Semantic wiki has been built (The Taaable system is accessible online at http://taaable.fr/). The Taaable system stores informal descriptions of cooking ingredients and recipes and an ontological structure about these items. End-users can update the text contents and the associated knowledge, and query the system to solve cooking problems. For example, a user can submit the following request: ”I want a dessert with rhubarb but without chocolate”. If no recipe exists with the specified characteristics, an existing recipe is adapted in order to answer the request. The system relies on a case-based reasoning (CBR) engine to perform adaptation. The CBR engine uses different data and knowledge sources: a set of indexed recipes, an ontology of ingredients, types of dishes, geographical origins of dishes and types of diets (vegetarian, nut-free, no alcohol) and a set of adaptation rules. The indexed recipes are computed from recipes written in natural language. An indexing tool uses the different ontologies to index the recipes. If the system is working fine, the maintenance of recipe base and knowledge is cumbersome. The indexing tool performs natural language processing which is error prone. In addition, the system is not able to capture the users feedback to improve its internal adaptation capabilities. Taaable has evolved in WikiTaaable in order to take advantage of collective intelligence. The knowledge base has been transfered into a semantic wiki with success. Users are now able to add recipes and to transform ontologies used by the Taaable system. However, the development of WikiTaaable raises several issues that are mainly related to the coherence of the system. How can we guarantee the coherence of the system while several users, often having different viewpoints, are allowed to modify the knowledge coded in the system? How can we efficiently combine a semi-automatic procedure and a manual enrichment process to build an ontology? These issues are of major importance because the ontology plays a central role in the Taaable system and is mainly used at two different levels. It guides the indexing process by identifying concepts involved in each recipe, and it is used by the CBR system to adapt recipes. If any user can freely modify the ontology, then the CBR engine and the recipe indexing bot might produce unpredictable results. These issues are clearly addressed by the Kolflow project.

Chosen methods and anticipated solutions The Kolflow partners have strong knowledge, experience and contacts in domains related to Kolflow issues. We already have tools for automated reasoning, tools for distributed collaborative systems and first experiences in man- machine collaboration. Partners also have data at each step of knowledge extraction process: target data, preprocessed data, transformed data, pattern and knowledge. We can experiment man-machine collaboration at each step of the process and check if some automated reasoners better support constraints of man-machine collaboration.

Work plan Man-machine collaboration scenarios must be ready at M6 and delivered to others activity. At the end of each year, this task will evaluate what parts of scenarios can be realized with advances performed by other activities.

Success criteria : The success of this is the measure of the success of the project. If the whole approach is working, then we should observe a better quality of ontologies with less overhead for the community in charge of knowledge maintenance. It means for the example of Taaable better user satisfaction for adaptation of recipes.

Risks : Man-machine collaboration raises complex issues. Maybe just some automated reason- ing can be transformed into smart agents.

Deliverables :

description Feb2011+months
D61 Man-machine collaboration scenarios 66
D62 Man-machine collaboration scenarios progress report 1212
D63 Man-machine collaboration scenarios progress report 2424
D64 Man-machine collaboration scenarios progress report 3636

Sub-tasks :

Participants :