Iterative application of a machine learning-based information extraction model to documents having unstructured text data

A computer program product comprising a non-transitory processor-readable storage medium having stored therein program code of one or more software programs, wherein the program code when executed by at least one processing device causes the at least one processing device to perform steps of:

• receiving a query to extract information from a document, the document comprising unstructured text data;
• performing two or more iterations of utilizing a machine learning-based information extraction model, wherein in each of the two or more iterations the machine learning-based information extraction model provides as output a portion of the unstructured text data extracted from the document and a relevance score associated with the portion of the unstructured text data extracted from the document in that iteration, wherein in a first one of the two or more iterations the machine learning-based information extraction model takes as input the query and the document, and wherein in subsequent ones of the two or more iterations the machine learning-based information extraction model takes as input the query and a modified version of the document with one or more portions of the unstructured text data of the document extracted in one or more previous ones of the two or more iterations removed therefrom;
• determining whether the portions of the unstructured text data extracted from the document in the two or more iterations have an associated relevance score exceeding a threshold relevance score and at least a threshold level of similarity to the query;
• and
• generating a response to the query, the response to the query comprising a subset of the portions of the unstructured text data extracted from the document in the two or more iterations determined to have associated relevance scores exceeding the threshold relevance score and at least the threshold level of similarity to the query.

A method comprising:

• receiving a query to extract information from a document, the document comprising unstructured text data;
• performing two or more iterations of utilizing a machine learning-based information extraction model, wherein in each of the two or more iterations the machine learning-based information extraction model provides as output a portion of the unstructured text data extracted from the document and a relevance score associated with the portion of the unstructured text data extracted from the document in that iteration, wherein in a first one of the two or more iterations the machine learning-based information extraction model takes as input the query and the document, and wherein in subsequent ones of the two or more iterations the machine learning-based information extraction model takes as input the query and a modified version of the document with one or more portions of the unstructured text data of the document extracted in one or more previous ones of the two or more iterations removed therefrom;
• determining whether the portions of the unstructured text data extracted from the document in the two or more iterations have an associated relevance score exceeding a threshold relevance score and at least a threshold level of similarity to the query;
• and
• generating a response to the query, the response to the query comprising a subset of the portions of the unstructured text data extracted from the document in the two or more iterations determined to have associated relevance scores exceeding the threshold relevance score and at least the threshold level of similarity to the query;
• wherein the method is performed by at least one processing device comprising a processor coupled to a memory.