Intent Detection via Ordered Sequence Feature Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing natural language processing systems fail to accurately capture user intent as an ordered sequence from user queries, missing essential components like desired actions and focal points, leading to irrelevant responses due to incomplete intent extraction.
Innovation Solution
A method and system that utilize word embedding features, Part of Speech tags, and dependency labels to create a feature vector for deep neural networks, enabling the detection of user intent as an ordered sequence, including desired actions and associated descriptors, by analyzing the semantic and syntactic significance of words in user queries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If user intent is captured as a single or contiguous sequence of text, then the extraction process is simple, but essential components like desired actions and focal points are missed
Solution Approach 1:
The patent segments user intent into distinct components: desired action, focal point, and descriptors. This is achieved through a neural network model that processes input sequences and outputs structured intent elements separately, ensuring each component is captured and processed independently for more accurate intent understanding.
2Ease of manufacture
If only the focal point is mined without the desired action, then the extraction is straightforward, but the mined intent becomes inaccurate and leads to unwanted results
Solution Approach 1:
The patent merges multiple intent components (desired action, focal point, descriptors) into a unified intent representation. The neural network model processes these combined elements together, allowing the system to capture the complete meaning of user queries by integrating all relevant intent aspects rather than treating them separately.
3Device complexity
If individual parts of user intent are extracted without capturing their interdependencies, then the extraction process is simpler, but the components do not make sense as a whole
Solution Approach 1:
The patent adds a temporal and structural dimension to intent extraction by processing components in a specific sequence and maintaining their hierarchical relationships. The neural network model outputs intent components in an ordered manner that reflects their interdependencies, allowing the system to understand not just what components exist but how they relate to each other in the context of user intent.
Data Source
AI summary
Disclosed herein is method and system for detecting intent as an ordered sequence from user query. The system identifies word embedding feature for each word using word embedding model and identifies Part of Speech tag feature. Thereafter, system determines dependency label feature based on dependency role and POS tag feature. Further, system provides feature vector comprising POS tag feature of target word, POS tag feature of previous two words of target word, word embedding feature of target word, word embedding feature of head word for each target word and dependency label feature of target word to deep neural network for detecting intent as ordered sequence. The ordered sequence includes desired action in user query, focal point pertaining to which desired action must be performed and one or more descriptors associated with focal point. In this manner, in present disclosure, overall intent from user query is captured for accurately providing response.