07-12, 11:00–11:10 (Europe/Amsterdam), Method (1.5)
We present TSearch, an experimental software system which attempts to unify the disparate, heterogeneous information sources across the Julia community into a coherent, searchable structure.
Online communities preserve their written knowledge across a diverse set of digital repositories. The Julia community, for instance, organizes Julia’s source code on GitHub; it conducts conversation on the Julia discourse; and it hold realtime conversations on Slack, and so on.
Despite the highly heterogeneous formats of these community repositories, they each encompass a unique mechanism of Julia related communication, with varying degrees of formality and structure. In the aggregate, these platforms reflect a web of people and their multitudinous ways of communicating.
Since the growth of the World Wide Web, search engines have primarily derived their ranking systems using a variety of techniques built from the graph structure provided by hypertext.
The supremely successful PageRank algorithm powered the explosive growth of Google by deriving rankings largely with hypertext signals.
We discuss the limitations that hypertext presents in any endeavor to derive complex structure, especially in the information ecosystems generated by a coherent network of people like the Julia community.
We present TSearch, an experimental attempt to unify these disparate, heterogeneous sources into a coherent, searchable structure. As our motivating application we use the Julia programming language, its extensive library ecosystem, and its diverse communication platforms.
Caleb Allen is a software engineer with experience developing applications and tools, primarily in startups. He especially enjoys building tools and infrastructure that make software development more enjoyable and productive.