CN-122003674-A - Generating search engine result documents

CN122003674ACN 122003674 ACN122003674 ACN 122003674ACN-122003674-A

Abstract

The present disclosure describes dynamically building and providing a generative search result document using a generative document system. The generative document system utilizes an aggregated framework that utilizes one or more Large Generative Models (LGMs). For example, the aggregation framework includes three phases, where a local process is applied to the generated output from the LGM, each phase built on the generated input from the previous phase. The generative document system uses an aggregated framework to create a generative search result document based on the search query and its corresponding search result links. These generated search result documents provide interactive, intuitive, comprehensive, and flexible choice of answers to process the corresponding search query.

Inventors

M. S.I. Gadit
A. J. Pagoto
S. S. Qiaoxi
E. R.T. abibo
N. J. Chalmers
SRIDHAR S

Assignees

微软技术许可有限责任公司

Dates

Publication Date: 20260508
Application Date: 20241110
Priority Date: 20231212

Claims (20)

1. A computer-implemented method for generating search result documents using a Large Generative Model (LGM), comprising: providing a generated document request hint (112) and a set of search result links (114) to a first LGM (116), wherein the set of search result links (114) is based on a search query (110); Receiving an unformatted generated search result document (118) from the first LGM (116), the unformatted generated search result document having a first topic portion (120) and a second topic portion (122), each of the first topic portion (120) and the second topic portion (122) associating a particular topic text narrative response with a search result link from the set of search result links (114); Receiving a first set of answer cards (130) from a second LGM (128) based on providing an answer card association prompt (126) to the second LGM (128), the first topic portion (120), and a set of candidate answer cards (124) associated with the first topic portion (120), wherein the first set of answer cards (130) associates a first text narrative response for the first topic portion (120) with answer cards within the set of candidate answer cards (124); Generating a first layout (138) for the first topic portion (120) using a first set of generated layout guides (136) generated by the third LGM (134) for the first topic portion (120) based on providing generated layout cues to the third LGM (134), wherein the first layout (138) visually locates the first text narrative response in combination with a first answer card from the first set of answer cards (130), and A formatted generated search result document (140) is generated that includes the first layout (138) for the first topic portion (120) and a second layout for the second topic portion (122).
2. The computer-implemented method of claim 1, wherein: Determining the first topic section and the second topic section are based on an analysis of the set of search result links; The first topic section includes the first text narrative response corresponding to a first subset of the search result links, and The second topic section includes a second text narrative response corresponding to a second subset of the search result links, the first subset being different from the second subset.
3. The computer-implemented method of claim 1, further comprising: Providing a first answer card query to the answer card system for the first topic section, and The set of candidate answer cards associated with the first topic portion is received from the answer card system.
4. The computer-implemented method of claim 1, wherein the first set of generated layout guidelines provides a ranked layout order between portions of text narrative responses in the first topic section and combinations of answer cards from the first set of answer cards.
5. The computer-implemented method of claim 1, wherein the first set of answer cards from the second LGM includes an available size of answer cards for each answer card within the first set of answer cards.
6. The computer-implemented method of claim 5, wherein: the first theme part is divided into grids with adjustable heights, and Generating the first layout is based on the available dimensions of the answer cards within the first set of answer cards.
7. The computer-implemented method of claim 1, further comprising providing the formatted generated search result document adjacent to the set of search result links to a client device in response to the search query, wherein the formatted generated search result document is generated based on the set of search result links.
8. The computer-implemented method of claim 6, further comprising modifying a ranking order of the set of search result links to prioritize the referenced search result links in the formatted generated search result document.
9. The computer-implemented method of claim 8, further comprising: storing the formatted generated search result document; Receiving additional instances of the search query; Comparing the set of additional search result links associated with the additional instance of the search query with the set of search result links associated with the search query, and Providing the formatted generated search result document from a storage device in response to the additional instance of the search query based on the additional set of search result links satisfying a relevance threshold with the set of search result links.
10. The computer-implemented method of claim 9, further comprising comparing the additional set of search result links to the set of search result links by: Selecting a plurality of search links as the relevance threshold, and The plurality of search links between the additional set of search result links and the set of search result links are matched in rank order.
11. The computer-implemented method of claim 8, further comprising: storing the formatted generated search result document; Receiving additional instances of the search query; determining that a ranking order of one or more links from a set of additional search result links associated with the additional instance of the search query does not match the search result links, and Based on the set of additional search result links, an updated formatted generated search result document is generated.
12. The computer-implemented method of claim 11, further comprising storing the formatted generated search result document, including storing a plurality of layouts for the first topic portion.
13. The computer-implemented method of claim 1, wherein at least two LGMs of the first LGM, the second LGM, and the third LGM are the same LGM.
14. A system for generating search result documents using a Large Generative Model (LGM), comprising: Processing system (900); and A computer memory (903) comprising instructions (905) that when executed by a processing system cause the system to: Receiving an unformatted generated search result document (118) from a first LGM (116), the unformatted generated search result document having a topic section with a specific topic text narrative response determined from a set of search result links (114) corresponding to a search query (110); Receiving a set of answer cards (130) for a topic section from the topic section in response to providing an answer card association prompt (126), a text narrative response from the topic section, and a set of candidate answer cards (124) associated with the topic section to a second LGM (128); generating a plurality of layouts for the topic section using a set of generated layout guides (136) generated by the third LGM (134) for the topic section based on providing generated layout cues to the third LGM (134), wherein the plurality of layouts visually locate the text narrative responses in combination with different answer cards from the set of answer cards (130), and A formatted generated search result document (140) is generated, the formatted generated search result document including a layout for the topic portion from the plurality of layouts.
15. The system of claim 14, wherein the operations further comprise: Receiving the search query from a client device associated with a user; Providing the search query to a search query system, and The set of search result links is received from the search query system.
16. The system of claim 15, wherein the operations further comprise: Providing the set of search result links to a fourth LGM having a search intent hint; receiving a set of potential search intents from the fourth LGM; selecting a search intent from the set of potential search intents, and The search intent, a generated document request hint, and the set of search result links are provided to the first LGM.
17. A computer-implemented method for generating search result documents using a Large Generative Model (LGM), comprising: Providing a generated document request hint (112) and a set of search result links (114) to the LGM, wherein the set of search result links (114) is based on the search query (110); receiving an unformatted generated search result document (118) from the LGM, the unformatted generated search result document having a first topic portion (120) and a second topic portion (122) each associating a particular topic text narrative response with a search result link from the set of search result links (114); Receiving a first set of answer cards (130) from the LGM based on providing an answer card association prompt (126) to the LGM, the first topic portion (120), and a set of candidate answer cards (124) associated with the first topic portion (120), wherein the first set of answer cards (130) associates a first text narrative response of the first topic portion (120) with answer cards within the set of candidate answer cards (124); Generating a first layout (138) for the first topic section (120) using a first set of generated layout guides (136) generated by the LGM for the first topic section (120) based on providing generated layout cues to the LGM, wherein the first layout (138) visually locates the first text narrative response in combination with a first answer card from the first set of answer cards (130), and A formatted generated search result document (140) is generated that includes the first layout (138) for the first topic portion (120) and a second layout generated for the second topic portion (122).
18. The computer-implemented method of claim 17, further comprising generating an additional layout for the first topic section using the first set of generated layout guides, wherein the additional layout visually locates the first text narrative response with a second combination of answer cards from the first set of answer cards, the second combination of answer cards being different from the first combination of answer cards.
19. The computer-implemented method of claim 18, further comprising: retrieving a stored version of the formatted generated search result document in response to receiving an additional instance of the search query; determining that an answer card in the first layout of the first topic section is not available, and In response to the additional instance of the search query, the additional layout for the first topic portion is provided to the formatted generated search result document.
20. The computer-implemented method of claim 18, further comprising generating an alternate layout for the first topic section, the alternate layout displaying the first text narrative response without any answer cards from the first set of answer cards.

Description

Generating search engine result documents Background In recent years, there have been significant advances in both hardware and software, particularly in the field of internet searching. Current systems are designed to retrieve and present a list of links and resources in response to a user search query. These systems also provide direct answers to frequently searched topics. However, as the number of accessible resources continues to increase, existing systems still have limitations in providing useful and relevant answers without requiring significant user effort. For example, users often have to manually examine the results list and view multiple entries before finding satisfactory answers. Sometimes, the user finds a direct answer to the search results, and the complete result list on the web page appears to be useless. These and other problems exist in current search result systems. Drawings The following detailed description provides specific and detailed embodiments with accompanying drawings. Further, each of the figures listed below corresponds to one or more embodiments discussed in the present disclosure. FIG. 1 illustrates an example overview of a generative document system that creates a generative search result document for a search query using a large generative model. FIG. 2 illustrates an example computing environment in which a generative document system is implemented. FIG. 3A illustrates an example sequence diagram of generating an unformatted generative document using a large generative model, and FIG. 3B illustrates an example of generating an unformatted generative document. Fig. 4A illustrates an example sequence diagram of determining answer card matches using a large generative model, and fig. 4B illustrates an example of generating answer card matches for a topic section. FIG. 5A illustrates an example sequence diagram of determining a layout guide using a large generative model, and FIG. 5B illustrates an example of generating a topic section layout. FIG. 6 illustrates an example sequence diagram of providing formatted generative documents for a previously searched query. FIG. 7 illustrates an example graphical user interface of a generative search result document. FIG. 8 illustrates an example sequence of actions for a computer-implemented method for generating a generative search result document using one or more large generative models. FIG. 9 illustrates example components included within a computer system for implementing a generative document system. Detailed Description The present disclosure describes dynamically building and providing a generative search result document using a generative document system that utilizes an aggregated framework that utilizes one or more Large Generative Models (LGMs). For example, the aggregation framework includes three phases, where a local process is applied to the generated output from the LGM, each phase built on the generated input from the previous phase. The generative document system uses an aggregated framework to create a generative search result document based on the search query and its corresponding search result links. These generated search result documents provide interactive, intuitive, extensive, and flexible choice of answers to process the corresponding search query. Embodiments of the present disclosure provide benefits and solve problems in the art by utilizing a system, computer-readable medium, and computer-implemented method for refining and building a generative document from search result links using one or more LGMs with a generative document system. As described below, the generative document system provides an aggregated framework that utilizes one or more LGMs to utilize different output types to construct a dynamic and interactive generative document to be provided with a search result link, wherein the generative document intelligently refines the search result information into a streamlined and understandable format. As mentioned above, the generative document system creates a generative search result document (generative document). To illustrate, in various embodiments, in response to receiving a search request from a user and obtaining a set of search result links, a generative document system generates a generative document request hint and a search result link and provides it to a first LGM that generates an unformatted generative document having a topic section with a text narrative response corresponding to a link in the search result link. In response to providing the answer card association prompt, the topic sections, and the candidate answer card sets for the topic sections to the second LGM, the generational document system receives a first answer card set or list for each topic section. In addition, the generative document system provides the generated layout hint and available layout options to the third LGM. In response, the generative document system receives a generative layout guide for eac