Snippet Search
No teams at the hackathon worked on this challenge, so a primary consideration would be to speculate on why it wasn't chosen. It was assumed that a simple challenge might have been a "gimme" for a team interested in doing something relatively straightforward, especially because a solution could have potentially be incorporated back into the open source Dspace software. Some speculations for lack of interest include:
- Maybe the project was too much like something one would make at work
- The challenge was not exploratory or creative or interesting enough, not really related to science per se
As it was discovered that there was a problem with text indexing on Dspace during the hackathon (because the DSpace API was not returning enough relevant or correct results), library staff spent time post-hackathon trying to fix DSpace indexing. As they solved that problem, they realized that some additional work on configuration of DSpace actually solved this challenge and provides the indexed text as snippets. Therefore a secondary learning would be to spend additional time on tech analysis of existing systems prior to the hackathon, testing thoroughly to make sure everything operates as it should (APIs, data, and system configuration overall).
- No projects created
Currently there is no easy way to do a full text search across one or more collections of digitized documents in our Digital Library (http://digitallibrary.amnh.org/handle/2246/6049) which uses DSpace as its platform. The full text search in DSpace provides some context but is incomplete and inconsistent. However, we currently have a website that solves this problem perfectly for one collection (Annual Reports). See - http://lbry-web-003.amnh.org/museum/annual_reports
Current annual reports search results interface
Unfortunately, the server that this site is running on will be taken down by end of year. We are not able to migrate it without additional expertise. Additionally, we’d like to have this granular search capability across multiple collections --not just Annual Reports.
We need a new interface that would allow us to search across library resources (including Annual Reports) in our DSpace Digital Library based on keywords and provide browsable “snippets” of information. The user can then choose to look at the full document(s) based on these “snippet views”. This is a more granular search than the “Dynamic Reference Search” challenge.
In terms of the existing interface, the "relevance" pie chart is not that relevant. The context of the keyword in the sentences of the search results is the most relevant aspect - but can you find more useful contexts, especially after examining the DSpace API? We're open to your ideas! One obvious filter would be to select or deselect which of the sub-collections in the research library are searched.
Our current Digital Library utilizes DSPACE (http://www.dspace.org/), an open source repository application widely used by Academic, for its platform. Consider creating solutions in the languages / platforms that DSpace is built in to be able to cleanly contribute back to their open source project which includes >1,000 Academic Institutions. Information on contributing to the DSpace Community can be found here: https://wiki.duraspace.org/display/DSPACE/How+to+Contribute+to+DSpace
- We want this tool to access the live versions of our Digital Library
- We want this to be created in a way that is extensible so we can add collections to our Digital Library that then can be added to this search.
- Display snippets of information with a link to expand
- User-friendly display (mobile-friendly?)
- Designed to be visually appealing, organizing a lot of information that is easy to read and navigate. Ideally a results page would display a list of relevant publications with a snippet of the search term in context for a user to browse without having to open a separate document.
- Provide facets to allow user to narrow search to one collection (ie Annual Reports) or multiple collections (ie Annual Reports and Natural History).
- http://lbry-web-003.amnh.org/museum/annual_reports/sorted?m=r&srch=boas
- DSpace Library Collections (http://digitallibrary.amnh.org/handle/2246/6049)
- DSpace REST API- https://wiki.duraspace.org/display/DSDOC6x/REST+API
- Ruby and Perl source code for current annual reports search application - on a flash drive, just ask for it!
