Named entity extraction is one of the subtasks of extracting information from unstructured text. To automate the process, the specialists label the words of certain groups, such as names, cities, addresses, and dates. For each task, we select the needed entities.

Named entity recognition (NER)

Go back

Technical assignment template

What goals can you achieve by Extracting Named Entities

SERVICES → LABELME.RU

The main task of NER labeling is to label certain classes in text data. For example, full name, locations and addresses, organizations and companies, dates and times, positions, phone numbers, emails, and so on. All this is necessary to automate the extraction, structuring, and understanding of texts.

Types of Named Entity Extraction

By types of output formats

There are various approaches to the allocation of named entities. The most common are BILOU, BIOES, and BIO. All of them serve as refinements to the main entities. In the case of BIOES, this is beginning, inside, ending, or single. Our specialists can work with all kinds of output formats. It ensures high annotation accuracy.

By different languages

It is crucial that the worker knows the language at the level of the native speaker and knows grammatical, lexical, and spelling standards. Otherwise, there may be miscalculations in the allocation of verbal constructions. Through local media and partner networks, we find specialists who are native speakers of the language. At the moment, LabelMe provides services for extracting named entities in English, Russian, French, and German.

By fields of application

In any field of activity, there are terms and nuances unknown to an ordinary person. It's especially relevant for technical or medical texts that contain specific data and entities. To minimize errors, we attract performers who have the necessary expertise and competence in the relevant field.

By the description of the beginning and end of the incoming phrase

Different NER tasks do not require different ways of allocating entities. In some cases, it is crucial to label one main word; in others — it is necessary to allocate whole constructions. The team of our performers easily adapts to the technical assignment features, marking what you need.

Application examples

The LabelMe team is working on data labeling for a wide variety of businesses:

Banks, accounting, and law

We have extensive experience in labeling a wide variety of documents for all key types of named entities: full names, locations and addresses, organizations and companies, dates and times, positions, phone numbers and emails. Our specialists will accurately extract all the entities you need.

Streaming services and information aggregators

Extraction of named entities is an important part of the work of recommendation or restrictive systems. For example, we labeled "forbidden words" in songs for a music streaming service. We considered all the client's wishes and expanded the technical assignment so that minors would not hear songs about death, murder, drugs, sex, etc.

Retail and E-commerce

We marked up tens of thousands of client requests, indicating the most important entities for further impact. It's necessary to both extract important order data and configure smart question-and-answer systems and chatbots.

Automation systems

There are many cloud services on the market for automating business and creating smart databases for enterprises. LabelMe adapts to the technical requirements and the needs of each client and creates datasets that eliminate the issues of doing business. We'll extract the classes of named entities you need from the data for processing, sorting, and structuring. We save both time and your money.

Leave a request