Serendipomatic architecture

Initially this page is for working through options for Zotero integration as input to the 'magic box' (the conceptual 'machine' that processes text and turns it into queries against various APIs), saving results to Zotero, multilingual text, etc, but it should eventually become further documentation for current and future developers.

Input into the magic moustache machine

Zotero integration as input to the 'magic box'

Related Issues:

• OAuth
• Handle "you're already in the database but another person is logged in" issue for Zotero

Inside the magic moustache machine

Parsing multilingual texts and turning it into queries

Related issues

• HTML tag stopwords
• Dealing with stop words and NER in multilingual texts

Query design

Related Issues:

• run api searches in parallel to avoid unnecessary slowdowns
• API query results should be mixed in order so the page is properly varied
• Need to limit the number of search results (also a results issue, but let's address it at the point of passing results to the screen first)

#Results from the magic moustache machine

Saving results to your Zotero library

Design constraints: not everyone has a Zotero account (or wants one). Some people use the browser plug-in, others use the standalone app. Zotero does not appear to work with COinS for all item types. API may be a better option. See https://forums.zotero.org/discussion/31226/some-item-type-wont-export-as-coins/ Handling public and private libraries.

Related Issues:

• Enhancing the Save Function