Building Acronym Tagging with Sequence Models for Many-to-Many Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for mapping building points represented by acronyms to standard tags fail to handle many-to-many relationships effectively, leading to inaccurate translations due to subjective naming conventions and lack of contextual understanding.
Innovation Solution
The implementation of a sequence-to-sequence neural network and statistical models, such as LSTM and CRF, that utilize contextual information to translate acronym strings into corresponding tags, enabling accurate handling of many-to-many mappings by considering the placement and relationships within the string.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a dictionary based mapping method is used to translate acronyms to tags, then the mapping process is simple and fast, but it fails to handle many-to-many relationships and produces inaccurate translations
Solution Approach 1:
The patent transforms the mapping approach from a static dictionary lookup to a dynamic sequence-to-sequence neural network model that processes contextual parameters. The model learns from training data to capture many-to-many relationships between acronyms and tags, significantly improving translation accuracy while maintaining reasonable processing speed through optimized inference.
Solution Approach 2:
The patent introduces contextual information as an intermediary element that mediates between the input acronym and the output tag. The sequence-to-sequence model processes the entire acronym string context rather than individual acronyms in isolation, enabling accurate disambiguation of many-to-many mappings by considering surrounding acronyms and their relationships.
2Measurement precision
If contextual information is incorporated to handle many-to-many relationships, then translation accuracy improves, but system complexity increases
Solution Approach 1:
The patent segments the complex translation task into two distinct phases: an encoder that processes the input acronym string and generates contextual representations, and a decoder that generates the output tag sequence. This segmentation of the sequence-to-sequence model makes the complex contextual processing manageable and enables accurate handling of many-to-many relationships through coordinated encoder-decoder operations.
3Adaptability or versatility
If standard tags are used instead of subjective acronyms, then system interoperability and energy optimization improve, but the mapping process becomes more challenging due to many-to-many relationships
Solution Approach 1:
The patent implements a feedback mechanism where the sequence-to-sequence neural network is trained on labeled data containing acronym-tag pairs. The model learns from this feedback to understand the many-to-many mapping relationships, continuously improving its translation accuracy. The trained model then provides reliable mappings that enable standard tag usage across building systems, achieving interoperability and energy optimization benefits.
Data Source
AI summary
A building system including one or more memory devices configured to store instructions that, when executed by one or more processors, cause the one or more processors to receive training data including acronym strings and tag strings, train a sequence to sequence neural network based on the training data, receive an acronym string for labeling, the acronym string comprising a particular plurality of acronyms, and generate a tag string for the acronym string with the sequence to sequence neural network, wherein the sequence to sequence neural network outputs a tag of the tag string for one acronym of the particular plurality of acronyms based on the one acronym and contextual information of the acronym string, wherein the contextual information includes other acronyms of the particular plurality of acronyms.