Support Center

Entity Extraction

Last Updated: Jan 23, 2015 10:26AM EST

Does Semantria service return entity type (person, company, title, etc.)?
Yes, for each entity responded through the document it will return with the type of entity.

Is there procedure to ingest user-defined sample files into user install?
A simple way to complete this task is through our Excel Add-In. Additionally, you can also send your list to to for ingestion.
Is there a way to predefine entity types in custom entity lists? 
Yes, you can do this through both the API and the Excel Add-In.
Custom (user) entities are specific things extracted from text which satisfies a user's defined pattern. In contrast to named entities which can be only places, people, companies, quotes, patterns, and job titles, custom entities can be anything the user wishes. It brings the user an ability to extract any custom nouns as entities with the sentiment and all the ensuing fields and options.

What are the named entity labels outputted?
Entity extraction returns output for the normalized entity title and the type of entity. (person, company, product, and more).
There are 2 “types” of entities, which are “named” and “user” entities. These values will appear in the “type” field.
Within each “type” (named and user), there are multiple options for the “entity_type”, which further classify the entity.
“Named” entities will have the following values:
·      Person
·      Company
·      Product
·      Place
·      Job Title
·      Regex or Pattern (date, time, phone number, zip code, address, twitter hashtag or handle, etc.)
“User” entities will have custom values, which the user has uploaded and assigned.
For more information on entity output, as well as other outputs in document processing mode please visit our developer’s portal.

What is the definition of custom entities and how does it work? 
Custom entities are supported by Semantria API 3.0. With this feature, customers are able to define the entity itself and its type, which may be useful for categorization purposes. For example, if you define the entity as a furniture type like this:
   “name” : “chair”,
   “type” : “furniture”
For the following text:
Most of the office chairs that are manufactured today come with twin wheel nylon casters. This type of caster, specifically the nylon wheel, can be used on carpeting but it will mark and scratch all wood floors. It happens because the nylon wheels don't roll, they slide. When they slide, they mark or scratch the floor. Our selection of hard wood chair casters is the ideal solution for this problem. The wheels on our chair casters for hardwood floors will not mark or scratch the floors.
Below are the results that are extracted from the above text.
   “normalized_form” : “chair”,
   “entity_type” : “furniture”,
   “type” : “user”
   “is_about” : true,
   “evidence” : 1.0
   “sentiment_score” : 0.675
So at the end it give us an excellent ability to seek for any entity we need not only for predefined ones such as Person, Company, etc.
seconds ago
a minute ago
minutes ago
an hour ago
hours ago
a day ago
days ago
Invalid characters found