Fig 1.
Diagram of current shortcomings in document-level chemical-disease relation extraction.
The left-hand text covers specific types of relationships between chemicals and diseases, in addition to related side effects and factors that influence these relationships. However, current research is limited to determining whether there is an association between chemicals and diseases.
Fig 2.
Flowcharts for precise relation extraction.
Fig 3.
Flowcharts for comprehensive relation extraction.
Fig 4.
Precise relation extraction results.
(a) the overall results of GPT 3.5, GPT 4.0, and Claude-opus in different workflows; (b) the bias results of GPT 3.5, GPT 4.0, and Claude-opus in different workflows; “a” represents Entity extraction, “b” represents Relation extraction, “c” represents Follow-up inquiry, and “d” represents semantic disambiguation.
Fig 5.
Comprehensive relation results.
(a) the overall results of GPT 3.5, GPT 4.0, and Claude-opus in different workflows; (b) the bias results of GPT 3.5, GPT 4.0, and Claude-opus in different workflows; “a” represents Main relation extraction, “b” represents Text structuring, “c” represents Side effects and condition extraction, and "d" represents Follow-up inquiry.
Fig 6.
The relationship between model extraction effect and entropy.
Fig 7.
The results of Statistical analysis for errors.
(a) the reasons for errors observed in GPT-4.0 data; (b) the factors contributing to incomplete extractions by GPT-4.0; (c) the reasons for errors observed in Claude-opus data; (d) the factors contributing to incomplete extractions by Claude-opus; “E -” represents errors in entity recognition; “R -” represents errors in relation recognition.