Search Tool for Customer Issue Sizing in Call Transcripts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional programmatic search tools are inefficient in processing large volumes of call transcripts and agent notes, leading to noisy and unreliable estimates of customer issues, slow search processing, and failure to identify relevant documents using synonyms or variations of search terms.
Innovation Solution
A method utilizing a processor to receive a query, process a corpus of documents by scoring relevance using a sparse document matrix, query vector, and optionally a similarity matrix and word coverage factor vector, to generate output identifying the most relevant documents based on sorting.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If conventional programmatic search tools are used to process call transcripts and agent notes, then the search process can be automated, but the search results become noisy and unreliable due to inability to accurately size customer issues
Solution Approach 1:
The patent transforms the search problem from exact term matching to similarity-based scoring by changing the parameters from binary (match/no match) to continuous (similarity scores). This allows the system to automatically process documents while maintaining accurate sizing through weighted similarity calculations that capture nuanced relationships between query terms and document content.
Solution Approach 2:
The patent replaces conventional mechanical search tools that rely on exact term matching with a computational system using matrix operations and similarity scoring. This substitution enables automated processing of large document corpora while producing reliable sizing estimates through mathematical similarity measurements rather than simple keyword searches.
2Reliability
If conventional programmatic search tools process each document for every search query, then complete search coverage is achieved, but the search processing becomes very time-consuming
Solution Approach 1:
The patent performs preliminary action by pre-computing the document term matrix and its transpose before actual search queries are executed. This preprocessing step creates reusable data structures that can be quickly applied to multiple queries without re-processing the entire document corpus each time, significantly reducing search processing time while maintaining complete search coverage.
Solution Approach 2:
The patent segments the search process into distinct computational stages: pre-computation of the document term matrix, construction of the query vector, matrix multiplication for scoring, and final result generation. This segmentation allows the system to perform expensive operations once during pre-computation and reuse the results across multiple queries, reducing overall processing time while maintaining reliability.
3Productivity
If conventional search tools only search for exact terms in the query, then the search process is simple and fast, but relevant documents using synonyms or variations of search terms are missed
Solution Approach 1:
The patent changes the search parameter from exact term matching to similarity-based matching by computing dot products between query vectors and document term vectors. This parameter change allows the system to quickly evaluate relevance while capturing synonyms and variations through the mathematical properties of vector similarity, preventing loss of relevant documents.
Solution Approach 2:
The patent creates a universal search mechanism using matrix operations that can handle both exact term matching and synonym/variation detection through the same computational framework. The document term matrix and query vector approach provides multi-functionality, enabling the system to retrieve relevant documents regardless of whether they use exact query terms or variations, while maintaining search efficiency.
Data Source
AI summary
The exemplary embodiments may provide a search tool that can locate customer issues in call transcripts and agent notes and can provide an accurate count of how often such issues appear in the call transcripts and agent notes. The exemplary embodiments may improve the speed with which the search of documents is performed. The exemplary embodiments rely upon a document matrix that is computed once for a given corpus of documents and a given vocabulary of the documents. The document matrix may be used across multiple queries. The exemplary embodiments also account for similar terms in processing a query. The exemplary embodiments may use a word coverage factor to improve the relevance of the search results returned by the search tool. The word coverage factor acts as a multiple factor that computes the fraction a query terms that are present in a document.