CE23 - Données, Connaissances, Big data, Contenus multimédias, Intelligence Artificielle

Modelling complex search tasks – CoST

Submission summary

While we consider information search nowadays to be 'natural' and 'easy', search systems are no longer able to provide adequate support for achieving a wide range of real-life work search tasks. In the CoST project, we particularly focus on the challenges underlying complex search tasks. The main characteristics of complex search tasks are the following: (1) the search spans over multiple sessions involving intensive user-system interactions with many cognitive resources involved; (2) the task is generally structured in multiple subtasks and/or multiple aspects or topics; (3) the task leads to dynamically linking and aggregating information across the user search history, and (4) the desired search outcome is the accomplishment of a task. Clearly, when achieving those tasks by means of search systems, users require support for task achievement that extends beyond the list of `ten blue links'. However, the state of the. art works reveal that such tasks are far from being solved today by traditional search systems
In the CoST project, we envision a shift from search engines to task completion engines by dynamically assisting users in making the optimal decisions empowering them to achieve multi-step and highly cognitive search tasks. This triggers the need for (1) more predictable and automatic models of user-system interactions and search tasks and, (2) more task-oriented information access models. The objectives envisioned in the CoST project are: (1) Identifying patterns of users' behaviours while completing complex search tasks. Our aim here is to discover behavioural regularities across users and relate them through clustering techniques that could explain the nature of the involved task; (2) learning explicit and structured representations of complex search tasks, based on those behavioural patterns. Our objective here is to capture the relationships and the dependencies between task stages, i.e., the overall structure of tasks; (3) modelling task-driven IR by relating document relevance to task completion. The driving idea here is to leverage from the search patterns on the one hand and the structure of tasks on the other hand, to establish possible user actions and rewards corresponding to the accomplishment of the task. The main scientific rupture intended in the CoST project is the definition of theoretical foundations of task-based IR, a radically new IR approach. This yields to: (1) the formal definition of a new concept, namely `task' viewed as a novel hidden variable to describe, characterize and incorporate in an IR model with respect to cross-fertilized views from computer science and cognitive psychology and (2) the design of new IR models that leverage from an operational task definition to shift the search success estimation from information relevance toward task accomplishment'. The general methodology relies on the use of deep learning approaches for user behaviour modelling and advanced information retrieval models based on reinforcement learning and personalization frameworks. More precisely, we investigate methods for identifying high-level task patterns from user-system interactions (WP1), which serve as inputs for learning the structured representations of complex search tasks (WP2) with the objective of formalizing retrieval models that optimize the search outcome (WP3). In addition, to leverage from the psychology cognitive view about search complexity a fourth work-package (WP4) is devoted for understanding and describing users' search strategies in this setting. Intensive experimental studies carried out within this work-package with real users will serve as the cognitive-based support for both learning and empirically validating the models and algorithms designed in WP1, WP2 and WP3. This fundamental research project is carried out by four partners specialized in information retrieval, machine learning, and cognitive science.

Project coordination

Lynda Tamine-Lechani (Institut de Recherche en Informatique de Toulouse)

The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.


IRIT Institut de Recherche en Informatique de Toulouse
LIG Patrick Levy
LIP6 Laboratoire d'informatique de Paris 6

Help of the ANR 544,463 euros
Beginning and duration of the scientific project: - 48 Months

Useful links

Explorez notre base de projets financés



ANR makes available its datasets on funded projects, click here to find more.

Sign up for the latest news:
Subscribe to our newsletter