IDCOLAB: An Interactive and Dynamic Collaborative Labeling Framework

Staff - Faculty of Informatics

Date: / -

USI Lugano Campus, room SI-006, Informatics building (Via G. Buffi 13)

You are cordially invited to attend the PhD Dissertation Defense of Saman KAMRAN on Tuesday, January 23rd 2018 at 15h30 in room SI-006 (Informatics building)


Collaboration relies on the flow of information among parties. To make informed decisions, shared knowledge should be managed in a way that each specific content becomes accessible to relevant group members.

One of the main factors that contribute to the quality of contents recommendations to the relevant users is the quality of identifying characteristics of the shared contents and users’ interests. A variety of automatic, semi-automatic or even manual methods exist for labeling and matching characteristics of the online contents and users’ interests. The low degree of Click Through Rate (CTR) on social platforms, however, suggests that labeling of shared contents and users’ active interests contains inaccuracies, leading to unsuccessful matching.

Although users’ perception about contents and their interests might change over time, most current approaches assign fixed characteristics to them because they postulate a fixed source of knowledge in their labeling process. Another common reason of unsuccessful matching is the heterogeneity of the labels assigned to the contents and users’ interest.

This dissertation introduces IDCOLAB, a novel framework that enables interactive and dynamic collaborative labeling which can result in more accurate matching of the users’ interests and shared contents based on the latest collective opinion of the evolving communities of users relevant to them. An essential step of IDCOLAB is the Semantic Augmentation Method (SAM) which enables collaborative labeling of shared contents by dynamically augmenting semantically related labels to labels initially assigned to the contents and users’ interests automatically or by an individual person. A goal of the augmentation process is to avoid irrelevant and noisy labels. In conjunction with this work, we have developed COD (Collaborative Ontology Development), which is a collaborative labeling platform for knowledge-sharing purposes at COD Technologies startup company. COD is a proof of concept model of IDCOLAB to evaluate our proposed framework in practice. We have evaluated IDCOLAB framework and SAM algorithm on COD platform with two separate focus groups in the domains of “Entrepreneurship” and “Artificial Intelligence”.

Dissertation Committee:

  • Prof. Mehdi Jazayeri, Università della Svizzera italiana, Switzerland (Research Advisor)
  • Prof. Fabio Crestani, Università della Svizzera italiana, Switzerland (Internal Member)
  • Prof. Marc Langheinrich, Università della Svizzera italiana, Switzerland (Internal Member)
  • Prof. Schahram Dustdar, Technical University of Vienna, Austria (External Member)
  • Dr. Thomas Tamisier, Luxembourg Institute of Science and Technology, Luxembourg (External Member)