Webhose.io, a fairly new data-as-a-service platform, has added a similarity search feature to the Webhose.io API. Similarity search allows users to search for articles that are similar to another document. Webhose.io has been busy adding new features and improvements to its API. Earlier this month, the company added viral post search capabilities to the Webhose.io API, and last month, it added named entity extraction capabilities.
The similarity search feature allows users to find articles based on a concept as opposed to verbatim keywords. The feature is particularly useful for public relations representatives and marketers who may want to keep track of the popularity of a specific press release and see how it is covered by media outlets and news blogs.
When an article is queried using the similarity search API, Webhose.io compares the title and text with other articles indexed by the platform. The frequency of uncommon terms present in the original article is then compared to other articles found in the Webhose.io index. Similarity search API calls should be made using POST instead of GET to avoid the maximum text length allowed with a GET request.
The API provides access to three days of history, so the similarity search feature works best with very recent articles. Articles that are older than three days will not return any results when queried using the API. Any type of article that quickly becomes popular and trends with the media should produce results with similarity search.
"The similarity search answers the needs that usually arise with PR companies and content marketers that need to find content, not by specific keywords, rather by a complete textual story," Webhose.io CTO and co-founder Ran Geva told ProgrammableWeb. "They need to monitor the performance and the coverage of a blog or a news article. Since online content may be rephrased, evolve or may even be completely rewritten, with the similarity search API you can still find it."
For more information about Webhose.io and the new similarity search API, visit the company website.