Deep Neural Network Token Model for Entity Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large-scale business applications face challenges in resolving ambiguities in entity or product information due to varying descriptions from different vendors and organizations, leading to customer confusion and reduced sales, especially in environments with large and rapidly changing item catalogs.
Innovation Solution
A deep neural network-based relationship analysis system employing an extensible multi-feature token model generates similarity scores by processing and comparing text attributes of item descriptors, using mirrored subnetworks and recurrent neural networks to determine the similarity or difference between items.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Stability of the object's composition
If standardized approaches are used to store item information, then data consistency is improved, but adaptability to different vendor formats deteriorates
Solution Approach 1:
The patent employs deep neural networks as an intermediary system that translates between different vendor-specific item description formats and a unified internal representation. The neural network model acts as a mediator that understands multiple vendor formats and converts them into consistent standardized data, resolving the contradiction between maintaining data consistency and adapting to various vendor formats.
2Measurement precision
If deep neural network with multi-feature token model is used, then measurement precision of item similarity is improved, but device complexity increases
Solution Approach 1:
The patent segments the item description analysis into multiple independent feature extraction components, including character-level features, word-level features, and attribute-level features. Each segment processes specific aspects of the description independently, and their results are combined to produce the final similarity score. This segmentation enables high measurement precision while managing system complexity through modular design.
Solution Approach 2:
The patent introduces multi-dimensional feature representation by analyzing item descriptions at multiple levels: character level, word level, and attribute level. This dimensional expansion allows the system to capture nuanced similarities that single-level analysis would miss, improving measurement precision while the hierarchical structure helps manage complexity.
Data Source
AI summary
A data set whose records include respective pairs of entity descriptors with at least some text and a representation of a relationship such as similarity between the entities of the pair is obtained. Using the data set, a neural network model is trained to generate relationship indicators for pairs of entity descriptors. In an extensible token model of the neural network model, a text token of a first attribute of a particular entity descriptor is represented by a plurality of features including a first feature which was added to the token model as a result of a programmatic request. A particular relationship indicator corresponding to a source entity descriptor and a target entity descriptor is generated using the trained neural network model.