NetOwl Extractor
Relationship Extraction
Relationship extraction is an advanced AI-based Natural Language Processing (NLP) technology that finds a variety of semantic relationships between entities in text. NetOwl provides a wide variety of relationship types applicable to many domains.
Advanced Relationship Extraction Powered by NLP
The power of entity extraction is that it can turn large volumes of unstructured text data into semantically-labelled structured information, which is key for many mission-critical applications like Intelligence Analysis, Fraud Detection, and E-Discovery. This should ideally include not only of names of entities, such as people, organizations, and places, but also the relationships among those entities, such as who is associated with whom, who owns what, who manufactures what, and where something is located.
While most entity extraction products offer only named entity recognition (NER) capabilities, NetOwl goes beyond NER to offer advanced NLP-based relationship extraction with an extensive relationship ontology pertaining to people, organizations, products, and places as well as state-of-the-art speed and accuracy.
Key Product Features
Accurate
Provides state-of-the-art relationship extraction accuracy even on noisy text.
Extensive Ontology Coverage
Extracts to a semantic ontology of over 40 pairwise relationships between entities.
Customizable
Creator Edition (CE) enables the customization of existing entity/relationship types or addition of new entity/relationship types.
Coreference Resolution
Resolves co-referring entities, whether they are names, pronouns, or definite noun phrases: “her” → “Amanda Smith”.
Fast & Scalable
Extremely fast for real-time analysis. Highly scalable entity and relationship extraction software with Docker and Kubernetes support.
Easy Integration
Easy-to-integrate relationship extraction product with a REST API. Pre-integrated with popular search and analytics tools like Elasticsearch and Esri ArcGIS.
The Challenges of Relationship Extraction
NetOwl handles the unique set of challenges that relationship extraction poses:
- Lexical Variation: A relationship can be expressed with multiple lexical choices:
- ABC’s employee/staff/representative Jim Stevens (employment relationship)
- Syntactic Variation: A relationship can be conveyed via different syntactic constructions:
- Jack Pettersen, an employee of XYZ Corp. (noun phrase) vs. Jack Pettersen works for XYZ Corp. (verb phrase)
- Coreference Resolution: Relationship extraction often involves a complex cross-sentence process called coreference resolution in order to figure out what entity a pronoun or definite noun phrase refers to. In the example below, Her needs to be resolved to Chris to extract the employment relationship and its needs to be resolved to ZZZ Co. to extract the headquarters-location relationship.
- Chris Jones is a sales executive who lives in a suburb of Chicago. Her employer, ZZZ Co., recently announced that it will move its headquarters to New York City.
Extensive Built-in Relationship Ontology
NetOwl’s relationship ontology offers a large set of pre-defined semantic relationships between many different types of entities, such as person, organization, place, product, address, phone number, etc., leveraging NetOwl’s broad entity extraction ontology.
Sample semantic relationships include affiliations between a person and an organization, associations among people, kinship relationships, various location-related relationships between a place on one hand and a person or an organization on the other, etc. NetOwl’s relationship ontology is useful in many domains such as Business, Politics, National Security, and Law Enforcement.
Additionally many important entity attributes, such as age and personal physical characteristics for a person or latitude/longitude values for a place (through Smart Geotagging), are also extracted, adding rich information to each extracted entity.
Coreference Resolution for Advanced Relationship Extraction
As discussed in the Why is Relationship Extraction Challenging? section, a relationship is often conveyed with a pronoun (e.g., her employer, its headquarters) or definite noun phrase (e.g., XYZ’s head of biomedical research). Successfully extracting a relationship in those cases requires 1) identifying not only named entities (e.g., Chris Jones) but also pronouns and noun phrases (i.e., the anaphors, in NLP parlance), and 2) identifying which named entities (i.e., the antecedents) the anaphors refer to. Since the antecedent entity is often mentioned in some earlier sentence along with many other entities, a complex process called coreference resolution is required that can weigh potential antecedents and choose the correct one.
NetOwl offers this advanced coreference resolution capability required for more complete and accurate Relationship Extraction.
Additionally many important attributes, such as age, personal physical characteristics, and latitude/longitude (through Smart Geotagging), are also extracted, adding rich information to each extracted entity.
Relationship Extraction for Link Discovery
Traditionally relationships between entities have been extracted manually by human analysts for use in a link analysis or visualization tool. With its automated relationship extraction, NetOwl can discover potentially useful new relationships from massive amounts of text data that could not have been thoroughly analyzed manually.
Analysts can then do what they do best, finding key insights and patterns from those newly discovered relationships, utilizing their deep subject matter expertise and extensive background knowledge. This type of human-AI collaboration can truly revolutionize knowledge discovery from Big Data.In addition to the more common search use case, another core use case for NetOwl NameMatcher is to compare two names to determine instantly if they represent the same name.
Relationship Extraction Solutions
E-discovery
Intelligence Analysis
Remove tag on Small
AML, KYC, PEP
Enterprise Search
Deploy on Premise or in the Cloud
Trusted by leading global organizations
Featured Blog Posts
What is Relationship Extraction?
Unstructured text contains a great deal of information about relationships between entities. Relationship Extraction finds them for you.
What Is Coreference Resolution?
And why is it hard? We walk you through a hallmark of Advanced Entity Extraction
Relationship Extraction is a Critical AI Technology for Effective Link Analysis
Relationship Extraction processes large quantities of unstructured data at scale to identify the critical links between entities.
Discover what NetOwl can do for you!
Schedule a demo
Request a no-cost evaluation
Request our whitepaper