Entity Set Expansion Through Multi-Method Ranking for Domain Adaptability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current natural language processing systems are limited in their ability to adapt to different domains and types of text corpora for entity set expansion, relying on manually curated data that quickly becomes obsolete and can only generate small knowledge graphs.
Innovation Solution
A system and method for entity set expansion that automates the process using both structured and unstructured data, employing multiple entity set expansion methods to determine associations and ranks, and combining these methods to generate a comprehensive entity set.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manually curated data is used for entity set expansion, then the system can generate entity sets with high initial quality, but the data quickly becomes obsolete and requires continuous manual updates
Solution Approach 1:
The system automatically evaluates and selects entity expansion methods based on performance metrics without requiring manual intervention. The processor executes multiple entity expansion methods, evaluates their performance on test entities, and automatically selects the best-performing methods for generating expanded entity sets, enabling the system to maintain high quality entity sets without continuous manual curation
Solution Approach 2:
The system dynamically changes parameters by selecting different entity expansion methods based on their evaluated performance. The processor adjusts which methods are used by changing the selection criteria parameters, choosing methods that perform best on specific types of entities and domains, thereby adapting to different data characteristics without manual reconfiguration
2Adaptability or versatility
If a single entity expansion method is used, then the system is simple to implement, but it cannot adapt to different domains and types of text corpus
Solution Approach 1:
The system achieves universality by implementing multiple entity expansion methods that can handle different domains and text corpus types. The processor executes a plurality of entity expansion methods including but not limited to relation extraction, event extraction, and entity linking, allowing the same system to adapt to various domains such as healthcare, finance, and news without requiring domain-specific customization
Solution Approach 2:
The system uses feedback mechanisms to evaluate the performance of each entity expansion method on test entities and automatically selects the best-performing methods. The processor calculates performance metrics such as precision, recall, and F1-score for each method, then uses this feedback to determine which methods to apply for generating expanded entity sets, enabling adaptive selection without increasing apparent system complexity
3Quantity of substance
If multiple entity expansion methods are evaluated and combined, then the system can improve recall and identify entities with different characteristics, but the processing time and computational resources increase
Solution Approach 1:
The system applies partial action by selectively evaluating entity expansion methods only on a test set of entities rather than processing all entities with all methods. The processor uses a portion of the data (test entities) to evaluate method performance, then applies only the best-performing methods to the full dataset, reducing overall processing time while still achieving improved entity set size through method combination
Solution Approach 2:
The system performs preliminary evaluation of entity expansion methods on a test set before applying them to the full dataset. The processor pre-assesses the performance of multiple methods using metrics like precision and recall on test entities, then selects the top-performing methods for actual entity set expansion, avoiding the computational cost of running all methods on all data
4Measurement precision
If the system is designed for a specific domain, then it achieves high performance on that domain, but it cannot be applied to other domains or text corpus types
Solution Approach 1:
The system is designed dynamically to adapt to different domains by automatically selecting entity expansion methods based on their performance on domain-specific test entities. The processor evaluates methods on test sets from different domains (e.g., healthcare, finance, news) and selects the methods that perform best for each domain, allowing the same system architecture to achieve high accuracy across multiple domains without manual reconfiguration
Data Source
AI summary
Disclosed embodiments relate to entity set expansion to associate with a text corpus. Techniques can include receiving unstructured data and a set of concepts associated with the data to determine, using a language model, a set of candidate entities in the data associated with the set of concepts, wherein the association is measured based on the relevancy of each candidate entity of the set of candidate entities to context of the data. Techniques can then determine, using a plurality of methods, associations between each candidate entity in the set of candidate entities and each concept in the concept of the set of concepts, wherein the each candidate entity is assigned a rank for each method of the plurality of methods. Techniques can use the assigned ranks to determine a combined rank of the each candidate entity of the set of candidate entities, wherein the combined rank of the each candidate entity is based on the assigned rank of the each candidate entity for the each method of the plurality of methods. Techniques can finally expand the entity set by determining a subset of entities of the set of candidate entities based on the combined rank of each candidate entity, wherein the subset of entities form the expanded entity set associated with the data.