A "Searchable" Space with Routes, for Querying Scientific Information

EasyChair Preprint 769, version 1

Abstract

Accuracy of scientific and technological information (STI) retrieval is on probation while its exploration expands: continuous additions of contents, comments and data structures create rich but uneasy conditions for selections of items by users of the same keywords. Moreover, users of information are confronted to a “hidden face” of searchable space: their own choices that could be recorded from their selections on lists of proposed answers to a query and could help mapping their navigation on recorded "routes" of search, are not open to consultation, and remain either unveiled or even unavailable. At best, users benefit from useful recommendations but, as data on their own preferences and choices is not shared, networked information for global navigation remains fuzzy.

In this position paper, our standpoint is  that search of scientific information generates data on various needs and users, and search would then take benefit from an efficient re-use of users  generated contents, in addition to any current metrics. First, with examples including our own data, we suggest that search practices in STI domains could be designed as an addition of sub-communities of users, sharing interest for their own selections of items. Second, we propose a data structure modeling interaction between those sub-communities and networking routes between them, in order to give a common room to "searchability" of a keyword domain. We thus propose a design of alternative paths to answers, which could better fit search architecture with observed needs for information retrieval in scientific research.

Keyphrases: Information Retrieval, Knowledge Management, Recommender Systems, complex systems applications, query and keyword modeling

@booklet{EasyChair:769,
  author    = {Renaud Fabre},
  title     = {A "Searchable" Space with Routes, for Querying Scientific Information},
  howpublished = {EasyChair Preprint 769},
  year      = {EasyChair, 2019}}