PoolParty Extractor

A graph-based text mining tool at the highest level

PoolParty_topics_text mining transparent

demos

The PoolParty Extractor is an intelligent text mining tool based in semantic technologies. The PoolParty Extractor analyzes documents and text, and extracts meaningful phrases, named entities, categories or other metadata automatically with high throughput and accuracy. Machine learning based Named-entity Recognition (NER) capabilities are also part of the PoolParty Extractor, allowing organizations to take advantage of natural language processing techniques and graph-based text mining.

The PoolParty Extractor is implemented in an “annotation pipeline,” where initial text mining is done through the Extractor, and additional services, such as Corpus Analysis and concept tagging, can be used to further enhance the annotations. This keeps the system flexible and allows it to be adapted quickly to new requirements.

Standout Features

Extraction of concepts, terms, and named entities

PoolParty Extractor provides a performant, secure and clusterable semantic extraction service (in the cloud or on-premise) that is able to automatically extract the most relevant concepts, terms, and named entities from a given document or text fragment.

this image shows an excerpt that is being extracted

Word sense disambiguation

One frequently observed challenge in text analytics is ambiguous terms, that is, when different concepts share the same label. This leads to incorrect annotations in the text extraction process. The PoolParty Extractor can distinguish such occurrences based on the taxonomy structure and the local context of the ambiguous concepts.

this image shows how the poolparty extractor can resolve language disambiguation

Specialized entity detection

The Extractor’s graph-based text mining capabilities recognizes special entities separately from others. There are built-in methods for recognizing regular expressions (e.g. product codes or file numbers), timestamps, names or locations.

this image shows how the poolparty extractor can detect special entities within a text

Classification and categorization

The Extractor can detect the “aboutness” of a text or text fragment through classification and categorization. Based on this methodology, domain-specific sentiment analysis we combine machine learning algorithms (SVM, Deep Learning, Naive Bayes, and more) with semantic knowledge graphs. The analysis is based on training calculations which are run in an easy-to-use training environment.

this image shows how the poolparty extractor can classify content

Corpus learning and scoring

PoolParty’s corpus analysis features include corpus scoring, which denotes a value to a term based not on frequency but instead its importance. The corpus analysis generates a list of terms from the reference corpora and using advanced algorithms, scores the terms that are most relevant to the text. The corpus scoring allows users to avoid incorporating general language in the taxonomy, and rather have specific concepts added to their domain.

the poolparty extractor has more precise methods for scoring extracted terms of a corpus

Main Benefits of PoolParty Extractor

- Advanced text mining based on SKOS standard taxonomies
- Ability to discern language ambiguities with concepts
- Contextual processing via semantic knowledge graphs
- Text analysis with natural language processing
- Enriched metadata for better findability of content
- Improved search and intelligent content recommendations
- Foundation for semantic knowledge models and applications

Semantic Solutions with Graph-Based Text Mining

The PoolParty Semantic Suite is celebrated for being easy to use and easy to integrate with existing customer systems. This works in favor of customers who wish to enhance their foundational taxonomies and text mining capabilities by building intelligent semantic solutions on top.

screenshot of poolparty extractor conducting an extraction of entities and concepts

Auto classification and concept tagging

Text mining in a thesaurus serves as the backbone for concept tagging. Concept tagging is a great resource for internal use in content management systems, where finding documents in a company drive is much faster because the documents have been tagged with precise authoring conventions.

Due to its APIs, PoolParty’s PowerTagging solution can be integrated into specific customer systems or used in ready-made integrations with SharePoint, Tridion Sites, and Adobe Experience Manager.

Semantic search

Using the concepts defined in the auto classification workflow, a semantic search engine can find products, documents, and other assets more precisely. If we consider our Puma sneakers example from before, a person in the UK might enter “black trainers” in the search field and get recommended products that have all been tagged under the concept “sneaker” because it has been bundled together with these synonyms and labels. Semantic search also employs a faceted approach, where by clicking on different facets (i.e. filters) that are modeled in the hierarchical structure of the taxonomy, customers can narrow down and tailor their results.

Semantic search capabilities in PoolParty PowerTagging for SharePoint.

See how in a matter of seconds, the number of search results were reduced from 250 entries to 14.

wine-cheese-recommender-system-poolparty-extractor

Recommendation capablities in free Wine & Cheese demo.

Taking the perspective of a deli vendor, see product profiles and their unique flavor pairings.

Personalized recommendations

PoolParty provides various similarity algorithms and services that exploit semantic knowledge graphs and/or extracted features of a document. A semantic recommender system can create “pairings” based on user activity or similarity of content, such as a ‘Show similar documents’ functionality while searching for relevant documents.

If a customer has just purchased a pair of black sneakers, for example, the recommender system will suggest a running tracksuit or other sports equipment – not more black sneakers. These queries are far more advanced and helpful to the user, fostering happy purchasing experiences.

How Customers Have Taken Advantage of Our Text Mining Tool

The Renewable Energy and Energy Efficiency Partnership (REEEP) is a customer of PoolParty who used text mining and the PoolParty Thesaurus to build their tool, the Climate Tagger. The Climate Tagger is a collection of documents and data pertaining to the climate industry that can be used as a reference point for researchers in the field.

More than 700 thousand documents were uploaded into PoolParty to extract and tag relevant concepts.

Read the story