Access and Recollection in Complex information SYStems (ARCSYS) – ARCSYS

In the context of Web interaction, memory and remembering are particularly important: individual pages are revisited and new pages are constantly added to the repertoire of visited pages. In order to revisit one page, we need to recollect what was on that page and also planning how to navigate to that page demands access to episodic memories of interaction with that web site. Retrieving information from memory represents a problem for web interaction, and older internet users may eventually show impaired ability to retrieve the context of previously seen items, their source, thereby leading eventually to a disorientation problem. Not only retrieving information from episodic memory is ecologically sound, because Web users commonly re-visit pages they have found earlier, but information re-access, the situation where the user found a piece of information earlier and now the same information is needed again, is one aspect missing from the existing models of information searching.

In this fundamental research project, partners share the assumption that emphasizing similarities between human memory and information retrieval can yield better cognitive models and better applications. Four series of experiments conducted by psychologists and ergonomists, computer scientists and information retrieval specialists are proposed, in which participants will interact with, access, remember, and eventually re-access contents. The contents consist in text(s) and are displayed by means of individual Web pages, a series of short documents on the same topic, bibliographic references accessed through an online reference management system (Connotea), a large collection of texts taken from the Text REtrieval Conference (TREC) series that will be retrieved with a flexible search engine.

The main goal of this research project is to develop a model of the long-term memory representations constructed as part of Internet users' navigation, access and processing of complex documents in digital information systems. Dual-memory processes have been described as implicit versus explicit, recollection versus familiarity, verbatim-based and gist-based judgements in the memory litterature. However, these retrieval processes have been neglected in prior research. Four series of experiments, or "Actions" are proposed ijn order to highlight their importance. (Action 1) Dual-process models are coordinated in order to derive predictions on the recollective experience and the nature of representations in memory in the context of Web interaction as a function of external factors (e.g. task-processing demands) and internal factors (e.g. aging). (Action 2) In source memory experiments, we compare memory for context and memory for content for Web pages and multiple texts. (Action 3) How visual search is controlled in interactive search tasks is analysed as a function of individual differences and saliency parameters. (Action 4) Finally, we explore the benefit of interventions designed for mitigating human-computer mismatches in relevance assessments and for improving re-findings activities in larger information retrieval environments with a social bookmarking platform and TREC collections.