Entity Extraction for Elasticsearch | Elasticsearch Named Entity Extraction

Quick and Easy Integration

Get better search results fast with Rosette’s 100% Java plug and play connection to Elastic. Download, install, and start querying with just one line of code.

Customizable

Rosette is pre-trained on 18 entity types and 20 languages and can be adapted in the field to domain-specific content for improved accuracy.

Cloud API

Easily deploy on the Cloud API and use it with the appropriate bindings for yourself. Sign up now to try.

Connect Your Unstructred Text to the Real World

Rosette not only extracts entities but links them to real-world knowledge base entries such as Wikidata. Go beyond entities and take the full advantage of Rosette with language identification, categorization, and sentiment analysis to get the most out of your unstructured text.

Visit our GitHub repository to learn more about the Rosette’s Document Enrichment plugin for Elasticsearch.
Get Started

This ingest plugin allows Elasticsearch users to perform Language Identification, Sentiment Analysis, Entity Extraction, Categorization, and Name Translation on documents as they’re indexed.

Frequently Asked Questions

What entity types can be returned by the /entities endpoint?

Rosette comes with 18 pre-trained entities such as person, organization, nationality, location, and so on. It can be adapted in the field to domain-specific content for improved accuracy. See the tech space section of our entity extraction capability page here to find the full list of pre-trained entities. Note that not all entity types are supported for all languages, please contact us for more information.

What are Rosette Entity Extraction and Entity Linking?

Rosette Entity Extraction SDK (REX) and Rosette Entity Linking SDK (RES) are available as Java SDKs and through the /entities and /entities/linked endpoints of the Rosette API (the /entities/linked endpoint is deprecated as of 7/2016 - use the /entities endpoint to get that same linking AND entity type return all from one endpoint) . Rosette Entity Extraction uses case-sensitive statistical models and exact matching to identify entities in documents in a variety of languages. Rosette Entity Linking links entities in the input to a knowledge base of people, locations, and organizations to help you establish the entity’s identity. The Rosette API endpoint returns an entityId that can be used to obtain more information about each entity, such as its canonical name, Wikidata page, and Freebase identifier. The Rosette Java SDK can be configured to use a user-managed entity database.

What kinds of text does Rosette Entity Extraction work with?

Rosette Entity Extraction was trained primarily on news articles. It works best with similar domains and can be adapted to perform well on text from other genres. However, it only works on sentences; not, for example, lists or spreadsheets.

See the Rosette API documentation for more information.

What's the difference between Rosette's Cloud API and On-Premise solutions?

Rosette is Basis Technology’s flagship product, a suite of linguistic tools that support a variety of different languages and can be accessed through multiple interfaces. When we refer to the “Rosette API” or our “Cloud API” we’re talking about the SaaS version of Rosette, a cloud-based RESTful web service that supports most of Rosette’s overall functionality. We host our API in the AWS cloud. Results are returned as JSON and we offer seven different client bindings plus a RapidMiner extension. You can sign up for a Rosette API account by visiting our developer portal. Just getting started? Sign up for a free plan and get up to 10,000 calls a month, no credit card required!

When we talk about Rosette On-Premise, we’re talking about versions of Rosette that are hosted by you and your organization. Using Rosette On-Premise gives you full control of your data and its security. In many cases, it also offers better latency than the Cloud API. Our On-Premise solutions also offer individualized customization options, including Rosette’s unique “state-ful” features — name indexing, custom entity extraction model training, and custom knowledge bases for entity linking — that are not available in the Cloud API.

Rosette On-Premise comes in two major “flavors” for different organizational needs:

Rosette API On-Premise: This version looks and feels like the basic Rosette Cloud API (all bindings and extensions are included), but is hosted on your private cloud or machine. You get to control the latency, and can decide which of Rosette’s functions and languages you want to utilize (but it comes with all of them on one install). Rosette API On-Premise comes with a web service RESTful API, as well as a simple Java API/JVM mode. All functions except Fuzzy Name Matching for ElasticSearch are available in this package.
SDK: This mostly-Java SDK that will look familiar to Java programmers and to current Basis customers, and will have the most flexibility and customizability for "state-ful" features. Some of our functions don't have specialized SDKs (like Sentiment Analysis and Relationship Extraction)— for those, Rosette API On-Premise is required. An SDK is especially appealing to users who want to just focus on one or two of our Rosette functions to a very in-depth level, and for users who are highly interested in our Name Indexing capabilities — our Fuzzy Name Matching for ElasticSearch runs on an SDK only.

If your organization has specific needs not addressed here, let us know! Rosette On-Premise can be customized to integrate into your system. Contact us for more information and a custom evaluation.