Abstract: A method includes obtaining a document with textual fields and a visual element. For each textual field, the method includes determining a textual offset for the textual field that indicates a location of the textual field relative to each other textual field in the document. The method includes detecting, using a machine learning vision model, the visual element and determining a visual element offset indicating a location of the visual element relative to each textual field in the document. The method includes assigning the visual element a visual element anchor token and inserting the visual element anchor token into the textual fields in an order based on the visual element offset and the respective textual offsets. The method also includes, after inserting the visual element anchor token, extracting, using a text-based extraction model, from the textual fields, structured entities representing the series of textual fields and the visual element.

Abstract: Systems, methods, and computer storage media having computer-executable instructions embodied thereon that facilitate providing suggested sites in accordance with target sites are provided. Web-browsed data is locally stored on a user device and uploaded to a suggestion engine. Additional web-browsed data is communicated to the suggestion engine via a suggested site request that provides a request for suggested sites related to a target site. In embodiments, the web-browsed data contained in the suggested site request includes data not previously uploaded to the suggestion engine. The suggestion engine utilizes the web-browsed data uploaded to the suggestion engine and the web-browsed data communicated via the suggested site request to provide suggested sites in association with the target site.

Abstract: Systems, methods, and computer storage media having computer-executable instructions embodied thereon that facilitate providing suggested sites in accordance with target sites are provided. Web-browsed data is locally stored on a user device and uploaded to a suggestion engine. Additional web-browsed data is communicated to the suggestion engine via a suggested site request that provides a request for suggested sites related to a target site. In embodiments, the web-browsed data contained in the suggested site request includes data not previously uploaded to the suggestion engine. The suggestion engine utilizes the web-browsed data uploaded to the suggestion engine and the web-browsed data communicated via the suggested site request to provide suggested sites in association with the target site.

Abstract: A finite state dictionary (200, 780) and method of production thereof by providing nodes (210–240) representing symbols stored in the dictionary; providing links (250–290) linking predetermined nodes; classifying each of the nodes; and associating with each node one of a plurality of formats selected in dependence on the node's classification. The node's classification is preferably representative of the node's probability of use. The plurality of predetermined formats preferably comprises: start of chain (210); heavy use (230); middle use (220); and light use (240). The probability of use of the nodes may be determined from a predetermined symbol probability analysis, or from a prior accumulation of statistics from use with corpora. This produces the advantage of increased run-time speed of dictionary look-up, and may reduce the dictionary's size.

Abstract: A method and arrangement for handling case and other orthographic variations in linguistic databases by explicit representation comprising: explicit storage of all orthographic and case variations of words in the dictionary, and use of extended cut and paste codes to control dictionary size explosion and to make the restoration of the lemma more efficient. This provides the advantage of allowing very efficient handling of case and orthographic variants while performing a dictionary lookup.

Abstract: A finite state dictionary (200, 780) and method of production thereof by providing nodes (210-240) representing symbols stored in the dictionary; providing links (250-290) linking predetermined nodes; classifying each of the nodes; and associating with each node one of a plurality of formats selected in dependence on the node's classification. The node's classification is preferably representative of the node's probability of use. The plurality of predetermined formats preferably comprises: start of chain (210); heavy use (230); middle use (220); and light use (240). The probability of use of the nodes may be determined from a predetermined symbol probability analysis, or from a prior accumulation of statistics from use with corpora. This produces the advantage of increased run-time speed of dictionary look-up, and may reduce the dictionary's size.