Keyword-Based Video Segment Creation Without Subtitles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for creating viewing segments of digital content based on keywords are limited, as they rely on subtitle information and struggle to handle content without subtitles or keywords not present in subtitles, making it difficult to search and segment content effectively.
Innovation Solution
A segment creation device and method that acquires character string data from multiple terminals, counts data matching a specified time zone and keyword, and creates segments based on consecutive time periods with a predetermined threshold value, allowing for keyword-based segmentation without relying on subtitle information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If subtitle information is used to decide the starting position of attention parts, then the search accuracy for content with subtitles is improved, but the applicability to content without subtitles deteriorates
Solution Approach 1:
The patent introduces a dual-path approach where subtitle information serves as an intermediary for content with subtitles, while audio analysis acts as an alternative intermediary for content without subtitles. The system selectively uses subtitle-based keyword matching when subtitles are available, and falls back to audio-based analysis when they are not, thus resolving the contradiction between search accuracy and adaptability.
Solution Approach 2:
The patent changes the parameter of information source from exclusively subtitle-based to multi-source (subtitle + audio). By adjusting the weight and usage of different information sources based on availability, the system maintains high search accuracy for subtitled content while gaining versatility to handle non-subtitled content effectively.
2Measurement precision
If subtitle information is used for keyword search, then the search capability for subtitle-present content is improved, but the search capability for keywords not in subtitles deteriorates
Solution Approach 1:
The patent makes the search system universal by enabling it to function effectively in two modes: subtitle-based keyword search for when subtitles are available, and audio-based keyword search for when they are not. The audio analysis component performs morphological analysis and keyword detection directly from audio data, providing the same search functionality regardless of subtitle availability.
Solution Approach 2:
The patent introduces audio analysis as an alternative intermediary between the search query and the content. Instead of relying solely on subtitle text as the intermediary, the system can use audio signal processing and speech recognition as another pathway to connect keywords with relevant content segments.
3Device complexity
If conventional segment creation methods are used, then the simplicity of the system is maintained, but the ability to create keyword-based segments without subtitles deteriorates
Solution Approach 1:
The patent segments the content analysis process into distinct modules: subtitle processing module, audio analysis module, keyword matching module, and segment creation module. This segmentation allows the system to maintain simplicity in each individual module while achieving complex keyword-based segmentation capability through their coordinated operation.
Solution Approach 2:
The patent makes the segment creation system dynamic by allowing it to adapt its processing path based on the presence or absence of subtitles. The system automatically switches between subtitle-based and audio-based analysis methods, creating a flexible and versatile segment creation capability without requiring a completely complex fixed architecture.
Data Source
AI summary
A segment creation device includes: a live data acquiring unit that acquires character string data transmitted from a plurality of terminals, and acquires live data that includes the character string data and a data registration date and time; a condition input unit that causes a user to input a search condition including time zone information representing a predetermined time zone and character string information; a live data number counting unit that counts the number of pieces of live data having character string data; and a segment creating unit that creates segment data which is specified based on a starting time and an ending time of a time period.