Combining computer implemented generalized intelligent document processing with personalized pathway selection

A system, comprising:

• at least one computer processor;
• a large language model (LLM), executing on the at least one computer processor;
• a personalized ranking model, executing on the at least one computer processor;
• and
• a document processing engine, executing on the at least one computer processor, and interfacing with the LLM, and the personalized ranking model, wherein the document processing engine performs operations comprising:
• classifying an uploaded document from a user application by the LLM to obtain a ranked list of document types of the uploaded document,
• retrieving a plurality of pathways corresponding to the ranked list of document types,
• executing a first portion of each of the plurality of pathways, using an extracted entity of the uploaded document as a parameter, to obtain a plurality of evidence structures, wherein the plurality of evidence structures corresponds to the plurality of pathways,
• processing the plurality of evidence structures using the personalized ranking model of a user of the user application, to obtain a ranked list of pathways,
• presenting, in the user application, the ranked list of pathways,
• receiving, from the user application, a selected pathway of the ranked list of pathways, and
• executing a second portion of the selected pathway.

A method, comprising:

• classifying an uploaded document from a user application by a large language model (LLM) to obtain a ranked list of document types of the uploaded document;
• retrieving, by a document processing engine, a plurality of pathways, each comprising an executable object, corresponding to the ranked list of document types;
• executing, in parallel, by the document processing engine, a first portion of each of the plurality of pathways, using extracted entities of the uploaded document as parameters, to obtain a corresponding plurality of evidence structures;
• processing the corresponding plurality of evidence structures using a personalized ranking model corresponding to a user of the user application, to obtain a ranked list of pathways;
• transmitting, by the document processing engine, the ranked list of pathways to the user application;
• receiving, from the user application, a selected pathway of the ranked list of pathways;
• executing, by the document processing engine, a second portion of the selected pathway;
• responsive to the selected pathway being a primary recommended pathway,
• updating a personalized vector of the personalized ranking model to increase weights of evidence features of the personalized vector, wherein an evidence structure corresponding to the selected pathway comprises the evidence features, and
• updating a historical match confidence score of the evidence structure;
• and
• responsive to the selected pathway being an alternative option,
• updating a personalized vector of the personalized ranking model to increase first weights of first evidence features of the personalized vector, wherein a first evidence structure corresponding to the selected pathway comprises the first evidence features, and
• updating the personalized vector of the personalized ranking model to decrease second weights of second evidence features of the personalized vector, wherein a second evidence structure corresponding to the primary recommended pathway comprises the second evidence features.