Mobile and Cultural Heritage, Medical IR, IR for E-tourism

Mobile and Cultural Heritage

Information access in mobility situation raises new challenges: Users are no more only concentrated on the search task, as the typical Web search engine usage. In mobility, the main task remains in the real world, like visiting a museum, being guide to seek for a good restaurant, etc. In this type of information access, context takes an important place: time, location, user activities, situation, etc. This need the studies of the integration of the user context and the current task into the IR Model. This research direction has being studied in the PhD of Mr. Kian Lam Tan in 2014. In his thesis, we propose a time and location depend IR Model, in which user interest and time are modeled using the physical analogy of capacitor. A mobile game on Malaysian Cultural Heritage has also being developed during this PhD.

One research direction studied by the MRIM team is related to usage of state of the art neural networks to support mobile interaction during museum visits. This work has been conducted in the context of the GUIMUITEIC FUI project. The goal of this project was to propose fully autonomous devices, due the museum constraints. The work of Maxime Portaz during his PhD defined solutions for two separated problems : 1) the recognition of masterpieces, and 2) the recognition of the gestures of the visitors in way to interact with the system. For both problems, the solution proposed is to define fully Convolutional neural networks (FCNN), specifically adapted to be used on mobile devices likes tablets. For the recognition of masterpieces, a FCNN that is able to use regions is proposed. One innovation of this proposal is to be able to learn the network without the need of regions input. The detection of user’s gestures is achieved through a global architecture composed of two main blocks (well adapted for mobile devices). This work has been published in the MTAP journal. This work was also used to participate to an international evaluation campaign in 2016.

Medical IR

Information retrieval in specialized domains requires to go beyond classical bag-of-words approaches. In the medical domain for instance, lay-users searching for health information may face difficulties expressing their information need or understanding documents retrieved. This issue, also known as the lexical gap, can be tackled by expanding the user queries with related terms, extracted from knowledge sources or embeddings built from medical data. Considering among the relevance criteria the understandability of documents can also be considered, as a way to lower the gap between the reader’s knowledge and the document. Learning to rank approaches allow to integrate such aspects in the IR process.

IR for E-tourism

This part is dedicated to the integration of web-based Information Systems. It aims at addressing issues that arise when designers want to assemble pieces of Information Systems accessible by the means of web services.

The studies involved have been conducted as a part of a broader project whose main goal is to design and implement a software system which provides context-aware personalized services for mobile users. This project was mainly funded by European Union’s Erasmus Mundus project “Sustainable e-tourism, Erasmus Mundus Action 2 programme”. More precisely, the issues addressed are related to mobile computing, and more specifically to system design, software architecture, expansion of information systems, distributed and heterogeneous resource access and integration (indexing and querying in highly distributed and heterogeneous environments), quality of service requirements (in presence of limited bandwidth and service discontinuity), and information synchronization when switching between connected and disconnected mode. Challenges we attempt to respond are those related to: i) new models for user’s privacy preservation in clustering and context aware recommendation (Mou Lei’s PhD Thesis, funded by the European Commission), ii) semantic composition of recommended services into a composite service, and execution of the resulting composite service (Pathathai Na Lumpoon’s PhD Thesis, funded by the European Commission), iii) discovery of services and recommendation (Isaac Caicedo’s PhD Thesis, supported by COLCIENCIAS–COLFUTURO Colombia scholarship, and also founded by the University of Cordoba in Colombia) and trip planning recommendation under constraints (Uyanga SUKHBAATAR Shukhbaatar, PhD Thesis, funded by the European Commission, joint degree with the national university of science and technology of Mongolia).

Laboratoire d'Informatique de Grenoble
Bâtiment IMAG
700 avenue Centrale
CS 40700, 38058 Grenoble Cedex 9 - France
Phone: +33 4 57 42 15 48
Group Leader: Georges QUÉNOT

*** MRIM M2 Internships proposals for 2022/2023 here***

Paper accepted for CIRCLE 2022 "Multi-element protocol for IR experiments stability. Application to the TREC-COVID test collection", with G. Nicole González Sáez, P. Mulhem, L. Goeuriot and P. Galuscakova.

Kodicare (continuous evaluation of web search engines) ANR International Research Project acccepted for funding. Cooperation with RSA Vienna and Qwant. 3 years, from 11/2019, finishing in september 2023.