Abstract: Architecture that enables the extraction of document-specific salient terms from documents for use improving the result summaries on a search engine result page (SERP), and methods to extract the salient terms from the documents using search engine logs, document metadata, and other algorithms. Document-specific salient terms can provide additional information and significantly improve user success in finding relevant documents while disregarding non-relevant documents. The architecture also enables the extraction of entity information from a variety of sources, some of which are at a query level, and other sources that are specific to a single document. All the entities available are aggregated for a set of results and the most relevant results are identified. The final set of results is then used to determine where in the document summary to apply visual emphasis or cues (e.g., bolding).

Abstract: Architecture that splits the generation of results summaries into at least two phases for reduced latency and improved relevance. A first phase generates a summary quickly and thereby enables subsequent modules to begin processing. A second round then executes in parallel to these other modules to offset the latency. The second round can also provide additional contextual information to the summarization module to improve the quality (relevance) of the summaries. Ultimately, the summaries generated in the two phases are merged. The multiple rounds of summarization enable the first round to be cheaper in processing resources to save overall latency, whereas the second phase can be more expensive, since the second phase can be run in parallel with other modules. Additionally, contextual information can be used to build the full content summaries in the second round.

Abstract: Architecture that splits the generation of results summaries into at least two phases for reduced latency and improved relevance. A first phase generates a summary quickly and thereby enables subsequent modules to begin processing. A second round then executes in parallel to these other modules to offset the latency. The second round can also provide additional contextual information to the summarization module to improve the quality (relevance) of the summaries. Ultimately, the summaries generated in the two phases are merged. The multiple rounds of summarization enable the first round to be cheaper in processing resources to save overall latency, whereas the second phase can be more expensive, since the second phase can be run in parallel with other modules. Additionally, contextual information can be used to build the full content summaries in the second round.

Abstract: Architecture that enables the extraction of document-specific salient terms from documents for use improving the result summaries on a search engine result page (SERP), and methods to extract the salient terms from the documents using search engine logs, document metadata, and other algorithms. Document-specific salient terms can provide additional information and significantly improve user success in finding relevant documents while disregarding non-relevant documents. The architecture also enables the extraction of entity information from a variety of sources, some of which are at a query level, and other sources that are specific to a single document. All the entities available are aggregated for a set of results and the most relevant results are identified. The final set of results is then used to determine where in the document summary to apply visual emphasis or cues (e.g., bolding).

Abstract: Search model updates are described. In one or more implementations, a search service uses a model to rank items in a search result, the model formed using an initial set of data. An update is generated using a subsequent set of data, which is formed after the initial set of data, that provides feedback describing user interaction with one or more items in the search result formed using the model, the generating of the update performed without use of at least a portion of the initial set of data. The model and the update are employed to rank items in a search result for a subsequent search.

Abstract: Search model updates are described. In one or more implementations, a search service uses a model to rank items in a search result, the model formed using an initial set of data. An update is generated using a subsequent set of data, which is formed after the initial set of data, that provides feedback describing user interaction with one or more items in the search result formed using the model, the generating of the update performed without use of at least a portion of the initial set of data. The model and the update are employed to rank items in a search result for a subsequent search.