Last week we introduced Enhanced Entities to News API, a comprehensive update of our entity enrichment and search features. This release includes a new entity recognition model (v3), enhanced search capabilities, and entity-level sentiment analysis to help you discover better quality data and business insights from our News API.
Enhanced Entities includes a number of improvements over our previous offering in News API:
Watch this short feature showcase video to discover the benefits of using Enhanced Entities.
Entity-based search vs Keyword search
Most of you will be familiar with the problems associated with keyword searches. They can take a long time to build, require frequent updates, and can still return a whole lot of noise. As a consequence, data scientists and engineers spend way too much time cleaning inaccurate data for models, and customers viewing news feeds inside apps or analysts sifting through search results are burdened with unmanageable noise and false positives. For example, one of the problems with keywords is that they don’t distinguish between very different things which have the same spelling, for example apple the fruit and Apple the company, or Zurich the city and Zurich the insurance group.
But you’ll be glad to hear there’s a better way: AYLIEN’s entity-based search. They take seconds to build rather than hours, and don’t require any maintenance or upkeep on your side, unlike those complex, hard-to-manage keyword searches that you’ve been relying on for too long. An entity is a real-world thing (a person, place, company etc) that’s mentioned in a news article and tagged with metadata by our News API. Our state of the art entity recognition model, backed by years of published NLP research, recognises over 5.6 million entities.
A practical example of Entity-based search in action
Imagine you are an analyst who covers risk events related to the insurance market. When searching for relevant and timely news about Zurich, a highly ambiguous word, you run the risk of being bombarded by far too much noise if you only search using the keyword Zurich. In fact this search will return hundreds of thousands of results incorporating everything from the city like its sports teams, tourism and much more. However, searching using a basic entity search for Zurich Insurance Group significantly reduces the number of articles returned, filtering out noise, and adding relevancy. You can refine results further using AYLIEN’s entity-level sentiment analysis to show just positive, neutral, or negative articles only.
Entity search isn’t just about cutting out noise though, it also returns all variations of an entity mention. Let’s examine MetLife as a second example. The keyword search will actually return slightly less results than an entity search. This is because an entity search includes multiple surface forms, not just MetLife, but Metropolitan Life Insurance Company, MLIC, and even the stock ticker MET (which you can also use to search). Whereas a keyword search will only return results for MetLife.
03 Jun, 2021
RADAR, AYLIEN’s new risk identification and monitoring solution, wins Deloitte Innovation Award
2 Min Read
02 Jun, 2021
News as a critical data source in Risk Monitoring and Management
13 Min Read
28 May, 2021
Advantages of a hybrid human/machine analyst research process
5 Min Read