Informal scholarly communication is an important aspect of discourse both within research communities and in dissemination and reuse of data and findings. Various tools exist that are designed to facilitate informal communication between researchers, such as social networking software, including those dedicated specifically for academics. Others make use of existing information sources, in particular structured information such as social network data (e.g. FOAF) or bibliographic data, in order to identify links between individuals; co-authorship, membership of the same organisation, attendance at the same conferences, and so forth.
Writeslike.us is a prototype designed to support the aim of establishing informal links between researchers. It makes use of data harvested from OAI repositories as an initial resource. This raises problems less evident in the use of more consistently structured data. The information extracted is filtered using a variety of processes to identify and benefit from systematic features in the data. Following this, the record is analysed for subject, author name, and full text link or source; this is spidered to extract full text, where available, to which is applied a formal metadata extraction package, extracting several relevant features ranging from document format to author email address/citations. The process is supported using data from Wikipedia.
Once available, this information may be explored using both graph and matrix-based approaches; we present a method based on spreading activation energy, and a similar mechanism based on cosine similarity metrics. A number of prototype interfaces/data access methods are described, along with relevant use cases, in this paper.
|Title of host publication||ELPUB 2010 - Publishing in the Networked World: Transforming the Nature of Communication|
|Number of pages||21|
|Publication status||Published - 2010|
|Event||14th International Conference on Electronic Publishing - Helsinki, Finland|
Duration: 16 Jun 2010 → 18 Jun 2010
|Conference||14th International Conference on Electronic Publishing|
|Period||16/06/10 → 18/06/10|