Deep search embedding of inferred document characteristics

A system for generating an output processed file, that is an electronic searchable file from an input original document, the system comprising:

• a computer processor;
• and
• memory coupled to the computer processor and storing instructions, when executed, causing the computer processor to:
• identify an object within the original document and a formatting attribute of the object;
• determine an inferred semantic characteristic of the object by comparing the formatting attribute with a plurality of inference rules, wherein the inferred semantic characteristic describes a purpose of the object in the original document and comprises an estimation of author-defined semantic characteristic of the object;
• detect location coordinates of the object in a visible layer of the original document;
• generate the processed file, where the processed file includes:
• a visible layer of the processed file that includes the object from the original document, where location coordinates of the object in the visible layer of the processed file correspond to the location coordinates of the object in the original document;
• and
• an invisible layer of the processed file that overlaps the visible layer of the processed file in a stacked format;
• generate metadata based at least on the inferred semantic characteristic and the formatting attribute of the object, wherein the metadata comprises text data that is searchable by a search application for the processed file to identify the object within the processed file;
• and
• add the metadata to the invisible layer of the processed file such that:
• location coordinates of the metadata in the invisible layer of the processed file correspond to the location coordinates of the object in the visible layer of the processed file;
• and
• the metadata in the invisible layer of the processed file overlaps the object in the visible layer of the processed file.

A non-transitory computer readable medium comprising instructions for generating an output processed file, that is an electronic searchable file from an input original document, the instructions, when executed, being configured to:

• identify an object within the original document and a formatting attribute of the object;
• determine an inferred semantic characteristic of the object by comparing the formatting attribute with a plurality of inference rules, wherein the inferred semantic characteristic describes a purpose of the object in the original document and comprises an estimation of author-defined semantic characteristic of the object;
• detect location coordinates of the object in a visible layer of the original document;
• generate the processed file, where the processed file includes:
• a visible layer of the processed file that includes the object from the original document, where location coordinates of the object in the visible layer of the processed file correspond to the location coordinates of the object in the original document;
• and
• an invisible layer of the processed file that overlaps the visible layer of the processed file in a stacked format;
• generate metadata based at least on the inferred semantic characteristic and the formatting attribute of the object, wherein the metadata comprises text data that is searchable by a search application for the processed file to identify the object within the processed file;
• and
• add the metadata to the invisible layer of the processed file such that:
• location coordinates of the metadata in the invisible layer of the processed file correspond to the location coordinates of the object in the visible layer of the processed file;
• and
• the metadata in the invisible layer of the processed file overlaps the object in the visible layer of the processed file.