Section: New Software and Platforms

PTraces

Personal digital Traces

Keywords: Information retrieval - Taxonomy induction - Personal Big Data - Ontology

Scientific Description Personal digital Traces (PTraces) is a platform for fetching and annotating personal data. Ptraces provides us a demonstration platform for personal semantics and ontology induction. Ptraces includes: (1) modules for securely fetching personal data from external applictions (2) an annotation component that annotates personal information with user-chosen taxonomies (facets). These facets arean experimental framework for testing the algorithms, (3) an information retrieval component

The PTraces software is composed of two parts: (i) A Java backbone, built on Elastic Search (ii) a web-browser user interface for configuring the system (choosing data sources, choosing ontologies to activate) and for query retrieved data.

Functional Description

The system accepts taxonomies in the SKOS format. These taxonomies are presented to the user, along with a number of connectors to otuside data sources (gmail, twitter, facebook, fitbit, ...). The user chooses which sources to index, and which taxonomies to apply. After this initialisation, the system accesses the sources using the indentification tokens that the user has supplied, fetches the user data, annotates the data using the activated taxonomies, and sets up a local server that that user can access to search in their own data, using a browser interface.

• Participants: Gregory Grefenstette, Lawrence Muchemi, Mohamed Bouatira

• Contact: Mohamed Bouatira

• URL: https://gforge.inria.fr/scm/browser.php?group_id=7217