A Study: From Electronic Laboratory Notebooks to Generated Queries for Literature Recommendation

A Study: From Electronic Laboratory Notebooks to Generated Queries for Literature Recommendation Oldooz Dianat, C´ecile Paris, and Stephen Wan CSIRO Computational Informatics Sydney, Australia firstname.lastname@csiro.au Abstract Relating one’s research to the vast body of scientific knowledge is a difficult task; the sheer volume of literature makes it diffi- cult to keep up-to-date with scientific de- velopments. Particularly when research is on-going, keeping track of related work is especially important to avoid an unin- tended duplication of effort. We outline a novel approach to this problem that uses the text in an Electronic Laboratory Note- book (ELN) as a representation of an ex- perimental context in the field of Chem- istry. The contribution of this work is to situate the literature recommendation task within the context of the user’s experi- mental information needs. We find that our approach to transform the ELN text into queries for use with PubMed is able to recover a subset of user bibliographies. We find that alternative methods for query generation that capture both scientific ter- minology and salient terms in the ELN complement each other. 1 Introduction Identifying the relationship between one’s re- search and the ever growing body of scientific knowledge is a time-consuming and laborious task. The sheer volume of existing literature makes it difficult to stay up-to-date with new sci- entific developments. Furthermore, this task is continual: this relationship must be revisited peri- odically so that one can avoid unintended overlaps with work published in parallel. To help keep up with advances by our scien- tific peers, we can use a number of tools to pro- vide continual exposure to newly published work. These can range from collaborative bibliography tools with a social network component, for exam- ple, the Mendeley application1. However, such tools do not have a mechanism to capture the in- formation needs of a researcher that might change on a daily or weekly basis due to the outcomes of experiments. In this work, we recognise the increasing use of Electronic Laboratory Notebooks (ELNs) in our research environments to capture a representation of research as it progresses. As part of the project described in this paper, we introduce the novel use of the text in the ELNs as a representation of the user’s context—specifically, their current experi- mental context—that provides insights on their in- formation needs. Our aim is to devise a system to transform this context into queries for a scien- tific literature search engine, and then suggest ref- erences that may be relevant. This paper describes the initial exploration in generating queries from the on-going experimen- tal context as represented in ELNs. In this work in progress, we investigate the effectiveness of differ- ent types of information extracted from ELN con- tent for the purposes of suggesting relevant litera- ture. The ability to suggest references in the context of reading an ELN entry is potentially useful in many contexts. Indeed, in our user study, we noted that users often mentioned the need to identify rel- evant literature based …

Content truncated to 500 words