Third Bridge: NLP for unstructured data

Challenge

Human-led insight, hard to search

Third Bridge is a market-leading global investment research provider of human-led insights for capital markets firms. It engaged RoZetta to automatically tag forum transcripts with companies from its reference data, so customers could search and discover relevant content more efficiently, and to lay a foundation for further enhancements.

Solution

Entity tagging at scale, then everything it unlocks

RoZetta’s data science experts and the DataHex platform mapped entities across 23,000 transcripts, from over 2.9 million entity mentions, averaging 125 per transcript.

The work linked to company fundamentals, news, and third-party and alternative unstructured sources, tracked entity sentiment over time, automated transcript summarization, tagged people, locations, and industries, and built an objective relevance measure for company mentions.

What DataHex delivered

Entity mapping across 23,000 transcripts Over 2.9 million entity mentions, averaging 125 per transcript.
Linking to fundamentals and third-party sources Company fundamentals, news, and alternative unstructured sources.
Entity sentiment tracked over time Plus automated transcript summarization.
People, locations and industries tagged With an objective relevance measure for company mentions.

Results

Discoverability from 37% to 98%

Over 152,000 entities tagged, ten times more than before
Discoverability of entity mentions lifted from 37.4% to 98.3%, creating a much better search and filtering experience
Automated summarization improved efficiency
Dataset linking surfaced relationships between entities
A sentiment index generated market-perception insight
An improved interface increased engagement, reduced attrition, and lifted revenue