Parsing implicit tables

A computer-implemented method of parsing implicit tables within a document and creating a trained database, the method comprising:

• collecting, by a processor, one or more documents in a document corpus, wherein the one or more documents in the document corpus exists in any format;
• creating, by the processor, a set of sentences that captures domain specific terminology in the document corpus based on a formalized writing style of the one or more documents in the document corpus to capture the domain-specific terminology from the document corpus via a text selector that is agnostic to a container mechanism of the one or more documents in the document corpus;
• training, by the processor, a neural network tailored for terminology specific to the document corpus using the set of sentences that captures domain specific terminology in the document corpus, wherein the set of sentences is vectorized and passed simultaneously in a trainable transformer structure including at least one of a Siamese and a triplet network architecture structure to derive semantically meaningful sentence embeddings that can be compared with cosine-similarity;
• parsing, by the processor, a document of the one or more documents from the document corpus to determine a pattern for a set of text that sets apart the information contained therein from the rest of the text;
• determining, by the processor, from the pattern of spacing, a first descriptive phrase and a second descriptive phrase;
• assigning, by the processor, the first descriptive phrase as an action and the second descriptive phrase as a trigger, wherein the trigger is a potential match for a user query and the action is a related task, and wherein the trigger is text against which the neural network compares the user query for the potential match and the action is a potential user action or topic, discovered by query processing by a neural network match between the user query and the trigger;
• adding, by the processor, the action and trigger to an action-trigger phrase set;
• accessing the neural network tailored for the document corpus;
• and
• creating, by the processor, the trained database by applying the neural network to the action-trigger phrase set, wherein the trained database is trained on the domain specific terminology and action-trigger phrase from the document corpus.