Information Processing Apparatus Multi-Input Sentence Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sentence generation techniques struggle to produce output data corresponding to multiple inputs, as they are designed to replace one sentence in a different domain, making it difficult to generate sentences from multiple inputs effectively.
Innovation Solution
An information processing apparatus that includes a selecting unit to abstract and identify a common concept among multiple inputs and a generating unit to create output data incorporating this common concept, allowing for the generation of sentences that summarize or represent multiple inputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If one sentence is replaced with another sentence in a different domain, then domain conversion is achieved, but the ability to generate output data corresponding to multiple inputs is lost
Solution Approach 1:
The patent segments the sentence processing into two distinct functional units: a selecting unit that handles domain conversion by selecting appropriate sentences from a database, and a generating unit that handles output data generation by creating new sentences based on multiple inputs. This segmentation allows each unit to specialize in one function, resolving the contradiction between domain conversion capability and multi-input processing capability.
Solution Approach 2:
The generating unit is designed with multi-functionality to handle both domain conversion tasks and multi-input processing tasks. It can generate output data corresponding to multiple inputs while also supporting domain conversion when needed. This universal design allows the system to maintain both capabilities simultaneously rather than having to choose one over the other.
2Extent of automation
If a learned model is used for machine learning based on text in target domain, then automatic sentence generation is achieved, but difficulty in generating output data corresponding to multiple inputs persists
Solution Approach 1:
The patent introduces an intermediary generating unit that acts as a mediator between the input sentences and the final output. This generating unit receives multiple input sentences, processes them through abstraction and concept identification, and generates the corresponding output data. This intermediary structure simplifies the overall system architecture by providing a dedicated component that handles the complexity of multi-input processing, making the automation more manageable and less complex.
3Measurement precision
If text replacing unit calculates probability information for word replacement, then precise domain conversion is achieved, but the system cannot handle multiple input sentences effectively
Solution Approach 1:
The patent segments the processing workflow into distinct stages: the selecting unit handles domain conversion with high precision through probability-based word replacement, while the generating unit handles multi-input processing to maintain productivity. This segmentation allows the precise word replacement mechanism to be applied only where needed for domain conversion, while other parts of the system focus on efficient multi-sentence processing, thus resolving the contradiction between precision and productivity.
Data Source
AI summary
An information processing apparatus includes: a selecting unit that abstracts a plurality of inputs based on the plurality of inputs and selects a common concept that is an abstract concept common to the plurality of inputs; and a generating unit that generates output data including the common concept based on a result of the selection by the selecting unit.