USER INTERACTION FRAMEWORK FOR TABEL, A DOMAIN INDEPENDENT FRAMEWORK FOR INFERRING SEMANTICS OF TABLES

Abstract

Data, structured in tabular form is omnipresent through all communication, research and analysis. Tables appear in print media, handwritten notes, computer software, white papers, architectural enhancement, and on the Web. Although the Web offers millions of tables, their interpretation is rarely evident to machines from the table itself. A novel way of automatically inferring the intended meaning of tables and representing it as RDF (Resource Description Framework) linked data has been implemented previously, but without user involvement through the process of interpretation. In this work we describe a user feedback framework to enable user involvement through the process of interpretation. We describe the design and implementation of a User Interaction Model, which allows the user to interact with the aforementioned system as a web service. The UI permits users to view and modify the system-generated interpretations for column headers and cell values and provide feedback if it is incorrect. We include a usability evaluation of the user interaction model to illustrate the effectiveness of the proposed model and subsequently discuss its role in improving the quality of the generated interpretations and in eliminating occasional misinterpretations.