Abstract: The disclosure describes a method including: obtaining a document image containing fields spatially arranged along multiple dimensions; determining first locations within the image, each of the first locations identifying a spatial position in the image of a respective one of the fields; extracting tokens from the fields; generating using a first trained model, based at least on the image, the first locations, and the tokens, first predictions, each first prediction indicating a classification of a respective one of the tokens; generating, using a second trained model, based at least on the image, second predictions identifying second locations, each of the second locations identifying a spatial position in the image of a respective entry of multiple entries, each of the entries including one or more rows of tokens; and linking, according to the first locations and the second locations, each of the tokens to one of the entries.

Abstract: Data cells in a spreadsheet report are collapsed and expanded without disrupting other spreadsheet data. A user may want to drill down on data corresponding to a parent member of the report to display more detailed information about the parent member. Likewise, a user may want to drill up the displayed information corresponding to the parent member to hide detailed information associated with the parent member. Before expanding or collapsing a report, the spreadsheet is scanned for information to determine whether the display of other data in the spreadsheet would be disrupted by the expansion/collapse. The information may include cross-join information associated with the parent member, the number of child members associated with the parent member, the dimension of the report created by the parent member and the associated child members, and other reports that may be linked to the report that includes the parent member.