Earlier this year, the Technology team at Quid participated in the second Quidathon. The two day event was particularly memorable as it was the last bit of innovation in our old office.
As with each Quidathon, there are only two rules: learn something new and have fun! Projects can range anywhere from implementing prototypes of new features, to exploring alternative technologies. We saw sixteen projects this time. Here are a few:
Visual Search: Explored an alternative workflow for our product by combining the search and visualization aspects together. Users were able to instantly generate a network from their search allowing for quick iteration and exploration.
Sharable Quid: Project began by providing the ability to share Quid layouts through embeddable images and eventually resulted in a fully interactive, annotated slideshow of insights users could share on blogs and social media sites.
Explicit Netgen: At Quid we excel at uncovering implicit relationships between unstructured text documents. The goal of this project was to explore the explicit relationships between entities. For instance, which investors have invested in the same companies and what does the corporate board membership landscape look like? The team implemented an explicit network generator which incorporated steps to prune connections according to statistical significance.
Quidding at a Distance: Pushing Quid out of the box. This team used websockets to explore what a potential multi-device Quid user experience would look like.
Optimized Data Store: Moved our graph storage from a SQL database to a NoSQL database. The time spent retrieving a 2,000 node network decreased from 45 seconds to 25 seconds (a 45% saving). Fetching a 6,000 node network went from 160 seconds to 120 seconds (a 25% saving).
Quid Knowledge Base: A central place for answers to common questions and troubleshooting tips for common issues.
Company Data Web Scraper: Built a web scraper to extract business descriptions from various websites with the hopes of improving Quid’s network generation for our company dataset.
Heatmap Layout: Added a heatmap layout to Quid’s visualization engine. The prototype leveraged D3 on top of Quid’s in-house engine to produce this alternative layout.
TL;DR Text Summarization: Built a extraction based text summarization algorithm that leverages document structure, sentence structure, key paragraphs, entities and keyword extraction to produce an apt summary of a text document.
The Quidathon was the perfect way to close out a chapter at our old office. We look forward to seeing how these ideas progress!
Interested in helping us solve awesome problems? If so, then head over to our careers page!