Microsoft Technology Licensing, LLC

Richard Critchlow

Lei Zhao

Lijiang Fang

Patrick H Halstead

Ashutosh Galande

Prakash Sundra Krishnamoorthy

Multilingual document searches are performed by ordering and filtering localized documents in different languages based on user preferences. Documents are indexed in a global search engine index using an identifier property and a language property permitting all relevant documents to be retrieved in a single search engine query. By storing documents with their identifier and language properties, documents from different cultures but with same content can be treated as equivalent. Results of the initial search include a multilingual set of documents ranked in order of relevance. Result sets are then post-processed using a data resource indicating for each multilingual user interface combination, which language version of each document should be returned. The initial search result are then transformed into a relevance based ordered list treating documents in different languages having same content as equivalent. Based on the ordered list, a final document list is prepared using localized and non-localized content.

Method and system for performing multilingual document searches

Search and filtering of documents with different metadata schemata is enabled using a single index that supports a single schema through decorated namespaces. Each metadata schema submitted to a system is assigned a unique identifier and property names associated with the schema are prefixed with the unique identifier. A single-valued, decorated property is used to indicate whether a submitted document is part of a registered schema in the system. The single-valued properties are converted to a search index table that enables resource-optimized searching and filtering of documents eliminating documents of other schemata by simply checking the association property.

Searching a document based on a customer defined metadata schemata

System and methods allowing for effective and reliable reading predictions for Japanese ideographs are provided. In an illustrative implementation, a reading predictions system operating in 'learning' and 'execution/run-time' modes is provided. In the &#x201c;learning&#x201d; mode the reading predictions system operates on a number of input sources to produce a decision tree that is used in the &#x201c;execution/run-time&#x201d; mode to return reading predictions for inputted Japanese sentences containing Japanese ideographs. Among the inputs utilized in the &#x201c;learning&#x201d; mode are base Japanese script readings, a training corpus, and quasi-phonological rules. From these inputs underlying readings and a decision tree are created. When operating in the &#x201c;execution/run-time&#x201d; mode, the reading predictions system employs a morphological analyzer to perform a morphology analysis on inputted sentences. Using the morphological analysis, the quasi-phonological rules, the underlying readings, and the decision tree reading predictions are provided.

Method for predicting the readings of japanese ideographs

Detecting typographical errors in a Japanese sentence by using a bottom-up approach analysis. The bottom-up analysis employs probabilities, dictionaries and heuristics to words that are found in morpho-lexical information derived from the Japanese sentence. This bottom-up approach combines valid phrases analyses into well-formed combined phrases, i.e., phrase lists, to determine the existence of &ldquo;holes&rdquo;. Holes are characters contained in the input sentence but not in the well-formed phrase lists. Probabilities are used to determine which phrase list is most representative of the input sentence. The hole contained in the phrase list having the lowest cost (highest probability) is analyzed to determine if it is a typographical error. This analysis includes checking the hole to determine if it is an extended dictionary and whether it is a proper noun. The hole may be &ldquo;relaxed&rdquo; by adding contiguous characters and rechecking the &ldquo;relaxed&rdquo; hole in the extended dictionary to determine if it is a proper noun. If the hole represents a typographical error, a replacement string is generated using reverse transformations to counteract the text entry error which created the typographical error. A dictionary is used in the replacement string generation process to determine the valid phrases.

Growing community of inventors

Seattle, WA, United States of America

Richard Critchlow