15+ min read
End-to-End Intelligence with AlphaSense Company Proximity Search
October 14, 2020
5 min read
Traditional search engines aren’t equipped to answer nuanced business queries, resulting in missed information and noisy results. AlphaSense addresses this pain with search algorithms tailored to extract the most valuable information from business and financial documents.
This week we launched company proximity search, a feature that takes the next leap in artificial intelligence-powered business research.
The release of company proximity search allows our search engine to leverage best-in-class company recognition and unique, innovative synonym technology to improve the breadth and quality of your search results. What does that mean? If you’re looking for trending themes from (J.P. Morgan) Chase, a traditional search engine will not understand the separation of Chase as a company, as a name, or as a verb. AlphaSense’s search engine understands the context in which these terms appear to deliver high accuracy statement-level company recognition. In addition, the search engine leverages the proximity of company mentions to the keywords you searched for to ensure that hits in research reports and news, which may discuss multiple companies, are, in fact, relevant to your search.
As it is generally conducted now, research can be a blocker to your ability to swiftly analyze a topic, market, company, or theme. So AlphaSense has invested nearly a decade toward research into an essential AI toolbox that applies natural language processing, search proximity, company tagging, bright synonyms, sentiment analysis, topic modeling, and knowledge extraction to optimize research for professional users.
Our AI Foundation
Our AI is built on a few core features that function beyond any box solution in the market:
- Smart Synonyms – An intelligent search feature that captures language variations by sourcing all relevant synonyms and applying them only in the correct contexts based on the intent of your search.
- Relevancy scoring – A multi-factor model that considers semantics, source, document structure, recency, and entity aboutness to deliver the most relevant documents to the top of the results list.
- Proximity – A method of incorporating the closeness between keywords into our determination of relevance.
- Sentiment analysis – A feature that enables users to identify, quantify, and analyze levels of emotion in human language. AlphaSense performs sentiment analysis at the phrase level to help sentiment search and aggregates those scores at the document and company level to enable macro trend analysis.
- Topic Modeling – A method of extracting and ranking the most important trending topics and themes affecting companies and industries. AlphaSense processes millions of documents in real-time to alert you to what’s happening.
- Knowledge Extraction – A strategy to extract structured knowledge from unstructured text like financial metrics discussed in earnings calls or press releases
- Company name recognition (or “CNR”) – a home-grown solution for recognition, inter-company disambiguation, and salience classification of company mention across AlphaSense content.
CNR paired with relevance scoring, smart synonyms, and proximity search leads to an AI that understands what or who you’re looking for in any given business context.
Our AI research team has already dedicated over four years of development and training towards improving the accuracy of company recognition technology alone. The beauty is in the subtlety of AI in research; you don’t even know it’s working if you aren’t thinking about it.
What is a company proximity search?
Company proximity search, built to maximize smart synonym, relevance, proximity, and CNR technology, is the next step in refining our already-powerful search engine. In addition, this product recognizes the difference between primary and non-primary mentions of companies to bring forward relevant data intelligently.
By leveraging company proximity search, you’ll uncover the most meaningful content and themes without limiting your search to company-published documents or title-level company mentions.
With company proximity search, you get all results sorted from most relevant to least relevant and still can be confident in a 3X reduction of noise while still isolating core themes. No other product in the market has attained sophistication for company recognition that can surface what other companies, researchers, and news sources are saying about any firm.
Company proximity search in your research
Have you been unable to recall important company information you’ve previously read because you did not remember the contents, the title, or which document it was from?
Our team has anticipated need and then gone beyond the normal AI-search functions to build a product that:
- Uncovers new peers and competitors, trace themes, and company correlation.
- Conducts concise competitive intelligence and instantly gets up to speed on core topics related to any company.
- Surfaces the macro impacts a piece has on multiple companies.
- Surfaces data on smaller companies with traditionally less Wall Street coverage.
- Runs company search across internal insights, equity research, industry reports, fixed income, and more.
Using company proximity search, you’ll be able to quickly isolate themes associated with any company within a variety of data sets. Of course, this kind of high-accuracy statement-level isolation can only be achieved through years of dedication and work.
More like this
14 min read
9 min read