17530867. GENERATION OF CAUSAL EXPLANATIONS FOR TEXT MODELS simplified abstract (International Business Machines Corporation)

From WikiPatents
Jump to navigation Jump to search

GENERATION OF CAUSAL EXPLANATIONS FOR TEXT MODELS

Organization Name

International Business Machines Corporation

Inventor(s)

Naveen Panwar of Bangalore (IN)

Deepak Vijaykeerthy of Bangalore (IN)

Nishtha Madaan of Gurgaon (IN)

Samiulla Zakir Hussain Shaikh of Bangalore (IN)

Diptikalyan Saha of Bangalore (IN)

GENERATION OF CAUSAL EXPLANATIONS FOR TEXT MODELS - A simplified explanation of the abstract

This abstract first appeared for US patent application 17530867 titled 'GENERATION OF CAUSAL EXPLANATIONS FOR TEXT MODELS

Simplified Explanation

The abstract describes a method for generating a causal explanation for the classification of a sentiment in an input sentence using a machine-learning model. The method involves splitting the input sentence into tokens, creating a causal subgraph based on the relationship between tokens, identifying the tokens that influence the classification, and generating a causal explanation that highlights the portion of the input sentence responsible for the classification.

  • The method receives an input sentence and classifies its sentiment using a machine-learning model.
  • The input sentence is split into tokens, where each token represents a term within the sentence.
  • A causal subgraph is created based on the causal relationship between tokens.
  • The causal subgraph is used to identify the tokens that have an influence on the classification.
  • A causal explanation is generated based on the identified tokens, highlighting the portion of the input sentence that led to the classification.

Potential Applications

  • Sentiment analysis: This method can be used in applications that require understanding the sentiment of text, such as social media monitoring or customer feedback analysis.
  • Natural language processing: The method can be applied in various natural language processing tasks, including text classification, sentiment analysis, and opinion mining.
  • Explainable AI: By generating causal explanations for classifications, the method can contribute to the development of explainable AI systems, which are important in domains where transparency and interpretability are crucial.

Problems Solved

  • Lack of interpretability: The method addresses the challenge of understanding the reasoning behind the classification of sentiment in an input sentence by providing a causal explanation.
  • Black box models: Many machine-learning models are considered black boxes, making it difficult to understand how they arrive at their decisions. This method helps to shed light on the decision-making process by identifying the influential tokens and generating a causal explanation.

Benefits

  • Improved transparency: The method provides a causal explanation that helps users understand why a particular sentiment classification was made, increasing the transparency of the machine-learning model.
  • Enhanced trust: By providing a clear explanation for the sentiment classification, users can have more confidence in the accuracy and reliability of the machine-learning model.
  • Insights into decision-making: The causal explanation generated by the method can provide valuable insights into the factors that contribute to sentiment classification, allowing for further analysis and improvement of the model.


Original Abstract Submitted

One embodiment provides a method, comprising: receiving an input sentence for a classification by a machine-learning model, where the classification is based upon a sentiment of the input sentence; splitting the input sentence into a plurality of tokens, each of the plurality of tokens corresponding to a term within the input sentence; creating a causal subgraph from the plurality of tokens, wherein the creating is based upon a causal relationship identified between tokens of the plurality of tokens; identifying, using the causal subgraph, tokens of the plurality of tokens influencing the classification; and generating, based upon the tokens of the plurality of tokens, a causal explanation for the classification, wherein the causal explanation identifies at least one portion of the input sentence resulting in the classification.

Retrieved from "http://wikipatents.org/index.php?title=17530867._GENERATION_OF_CAUSAL_EXPLANATIONS_FOR_TEXT_MODELS_simplified_abstract_(International_Business_Machines_Corporation)&oldid=25139"