Hierarchical Taxonomy for Comprehension-Based Question Answering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current pre-trained language models for question answering lack true comprehension and often memorize information without deeper understanding, failing to effectively utilize knowledge for specific tasks.
Innovation Solution
Implementing a hierarchical knowledge taxonomy that uses a pre-trained language model to respond to questions by selecting layers with varying complexity, limiting responses to words associated with a layer of complexity one level less than the question, to promote comprehension-based question answering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If pre-trained language models are used to answer questions, then the models have extensive knowledge, but they fail to truly understand the content and simply memorize without deeper comprehension
Solution Approach 1:
The vocabulary is segmented into multiple layers based on complexity levels (Layer 1: simplest words, Layer 2: intermediate words, Layer 3: most complex words). The question and answer vocabulary are separately segmented, and the model is trained to map questions from one layer to answers in a different layer, preventing memorization by enforcing structural transformation rather than direct recall.
Solution Approach 2:
Different parts of the vocabulary space are assigned different qualities (complexity levels). By requiring the model to transform between specific local regions of the vocabulary space (e.g., Layer 3 question to Layer 1 answer), the system ensures that comprehension is demonstrated through appropriate complexity transformation rather than simple repetition.
2Adaptability or versatility
If the language model uses all available words to respond to questions, then the model has access to maximum knowledge, but the model cannot demonstrate true understanding and comprehension
Solution Approach 1:
The system dynamically adjusts which vocabulary layers are used for questions versus answers based on the desired complexity transformation. The model can be configured to transform from high-complexity questions to low-complexity answers or vice versa, making the comprehension task adaptive rather than static. This dynamic layer assignment prevents the model from simply recalling memorized patterns.
3Reliability
If the hierarchical taxonomy with multiple complexity layers is implemented, then comprehension-based question answering is improved, but the system complexity increases
Solution Approach 1:
The vocabulary space is extended into a new dimension (complexity layer) rather than simply adding more words to a flat vocabulary. This dimensional organization allows the model to demonstrate comprehension by navigating across layers, and the layer assignments can be determined automatically using existing complexity metrics, reducing manual curation complexity.
Data Source
AI summary
A method, apparatus and system for comprehension-based question answering using a hierarchical taxonomy include receiving a word-based question, associating the word-based question with a layer of the hierarchical taxonomy, in which the hierarchical taxonomy includes at least two layers, each of the at least two layers including respective words resulting in the at least two layers having varying levels complexity, determining which layer of the at least two layers of the hierarchical taxonomy comprises a layer of complexity one level less than the layer of the hierarchical taxonomy associated with the word-based question, and using a pre-trained language model, answering the word-based question using only words associated with the layer of the at least two layers of the hierarchical taxonomy having the one less level of complexity.