Abstract: Operations of a search management system are disclosed. The operations may include: identifying a data corpus containing a plurality of documents, generating sets of feature vectors representing the plurality of documents, receiving a query to search the data corpus, generating a query vector for the query, identifying a target feature vector that meets a similarity threshold by comparing the query vector to the feature vectors, and presenting a query result that includes at least part of the document. The feature vectors may be generated by executing a multi-step partitioning process for partitioning a respective document into a plurality of document partitions, such that the sets of feature vectors that are generated correspond to the plurality of document partitions for the respective document. The query result may include a target partition from among the plurality of document partitions represented by the target feature vector.

Abstract: Operations of a search management system are disclosed. The operations may include: identifying a data corpus containing a plurality of documents, generating sets of feature vectors representing the plurality of documents, receiving a query to search the data corpus, generating a query vector for the query, identifying a target feature vector that meets a similarity threshold by comparing the query vector to the feature vectors, and presenting a query result that includes at least part of the document. The feature vectors may be generated by executing a multi-step partitioning process for partitioning a respective document into a plurality of document partitions, such that the sets of feature vectors that are generated correspond to the plurality of document partitions for the respective document. The query result may include a target partition from among the plurality of document partitions represented by the target feature vector.

Abstract: Operations of a search management system are disclosed. The operations may include: identifying a data corpus containing a plurality of documents, generating sets of feature vectors representing the plurality of documents, receiving a query to search the data corpus, generating a query vector for the query, identifying a target feature vector that meets a similarity threshold by comparing the query vector to the feature vectors, and presenting a query result that includes at least part of the document. The feature vectors may be generated by executing a multi-step partitioning process for partitioning a respective document into a plurality of document partitions, such that the sets of feature vectors that are generated correspond to the plurality of document partitions for the respective document. The query result may include a target partition from among the plurality of document partitions represented by the target feature vector.

Abstract: A document may be received as part of a request to identify similar documents in a collection of documents. However, the received document and the documents in the collection may have different schemas or formats. To provide semantic context to the search and allow similarity scores to be generated between different document types, a configuration may be accessed that defines how to generate queries from one schema into another schema. The configuration may map queries between different fields in both schemas. Results of the multiple queries can be combined to generate a weighted combination for each document that can be used as a similarity score between different document types.

Abstract: A document repository may be searched for documents that are similar to a source document. Multiple queries may be generated based on a type of the source document, and the results may be combined in a unified response. User behavior may then be monitored, and implicit and explicit feedback may be gathered to evaluate the performance of the search. The gathered feedback may indicate how relevant each of the result documents are in comparison to the original source document. This feedback may then be used to adjust search parameters for the source document type, such that the performance of subsequent searches may be improved. A model may also be trained to classify implicit feedback using explicit feedback received from users.