Subsymbolic Encoder Preserving Semantic Relationships

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing AI systems face challenges in combining symbolic and subsymbolic data approaches, as conversion methods like the bag-of-words model often lose syntactic and semantic information, preventing effective reproduction of semantic relationships between symbols.

Innovation Solution

A subsymbolic encoder system that converts symbolic data into a numeric vector representation while preserving syntactic and semantic information by using an exponential component to maintain the ordering and relationships of symbolic elements, allowing for the combination of symbolic and subsymbolic data processing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If symbolic data is converted into subsymbolic form using the bag-of-words model, then the data can be processed by subsymbolic AI systems, but syntactic and semantic information is lost

Engineering Contradiction:

Improvecompatibility with subsymbolic AI systemsVSAvoidsyntactic and semantic information

Core Design Contradiction:

Adaptability or versatilityVSLoss of information

Solution Approach 1:

The patent transforms symbolic data into subsymbolic vector representations by changing the parameter representation from discrete symbols to continuous vectors, while preserving semantic relationships through carefully designed vector operations that maintain the structural parameters of the original symbolic data

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent introduces an intermediary encoding system that acts as a bridge between symbolic and subsymbolic representations. This intermediary layer uses vector operations to translate symbolic relationships into subsymbolic form without losing essential semantic information, enabling communication between the two AI paradigms

Inventive Principle:

Principle #24Intermediary (Mediator)

2Ease of operation

If symbolic AI systems use human-readable representations, then the data is interpretable, but processing speed is slower compared to subsymbolic AI

Engineering Contradiction:

Improveinterpretability of dataVSAvoidprocessing speed

Core Design Contradiction:

Ease of operationVSProductivity

Solution Approach 1:

The patent segments the processing pipeline into distinct stages: symbolic data preparation, vector encoding transformation, subsymbolic processing, and result decoding. This segmentation allows each stage to be optimized independently, maintaining interpretability in symbolic stages while achieving high-speed processing in subsymbolic stages

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent implements dynamic switching between symbolic and subsymbolic representations based on the processing requirements of different data types and operations. This allows the system to maintain interpretability when needed while achieving high-speed processing when appropriate

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS11288581B2Subsymbolic encoding with preserved semantic relationships

Publication Date: 2022.03.29 SAP SE

AI summary

Disclosed herein are system, method, and computer program product embodiments for encoding symbolic data into a subsymbolic format while preserving the semantic arrangement of the symbolic data. In an embodiment, to encode the symbolic data, a subsymbolic encoder system may convert a symbolic graph into a tuple representation having tuple elements corresponding to the nodes of the symbolic graph. The subsymbolic encoder system may retrieve a dictionary identification for each tuple element and calculate a subsymbolic value for each tuple element using an exponential component. The subsymbolic encoder system may standardize the length of the subsymbolic values and/or add a weighted relationship indicator to the subsymbolic values. The subsymbolic encoder system may transmit the subsymbolic values to a subsymbolic intelligence system.