WO2018133648A1 - 搜索方法、装置及非临时性计算机可读存储介质 - Google Patents

搜索方法、装置及非临时性计算机可读存储介质 Download PDF

Info

Publication number
WO2018133648A1
WO2018133648A1 PCT/CN2017/119820 CN2017119820W WO2018133648A1 WO 2018133648 A1 WO2018133648 A1 WO 2018133648A1 CN 2017119820 W CN2017119820 W CN 2017119820W WO 2018133648 A1 WO2018133648 A1 WO 2018133648A1
Authority
WO
WIPO (PCT)
Prior art keywords
landmark
search
cluster
materials
clustering
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2017/119820
Other languages
English (en)
French (fr)
Inventor
杨荣权
覃婷立
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Sankuai Online Technology Co Ltd
Original Assignee
Beijing Sankuai Online Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Sankuai Online Technology Co Ltd filed Critical Beijing Sankuai Online Technology Co Ltd
Priority to CA3078148A priority Critical patent/CA3078148C/en
Publication of WO2018133648A1 publication Critical patent/WO2018133648A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33Querying
    • G06F16/3331Query processing
    • G06F16/334Query execution
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/35Clustering; Classification
    • G06F16/355Creation or modification of classes or clusters

Definitions

  • the present application relates to the field of search technology, and in particular, to a search method, apparatus, and non-transitory computer readable storage medium.
  • a query word can be obtained, and the search engine first determines the user's search intention according to the query word, and then selects a corresponding search strategy to perform a search operation according to the user's search intention.
  • the present application provides a search method that solves the problem of inaccurate search results caused by the inability to identify search intents matching the query words existing in the prior art.
  • an embodiment of the present application provides a search method, including:
  • a matching third landmark material is recalled based on the search intent.
  • an embodiment of the present application provides a search apparatus, including:
  • the non-transitory computer readable storage medium stores machine executable instructions executable by the processor, the processor being caused by the machine executable instructions:
  • a matching third landmark material is recalled based on the search intent.
  • embodiments of the present application provide a non-transitory computer readable storage medium storing machine executable instructions that, when invoked and executed by a processor of a network device, cause the processing Execution:
  • a matching third landmark material is recalled based on the search intent.
  • the search method disclosed in the embodiment of the present application matches the obtained query words with the landmark materials in the preset landmark database, and then clusters the landmark materials that are successfully matched based on the distance between the landmark materials, according to the clustering. As a result, the correlation of the landmark material that is successfully matched is determined. If the correlation is greater than a preset threshold, the user's search intention is determined to be a landmark search, and the material recall is further performed according to the clustering result, thereby solving the existence of the prior art. The problem of inaccurate search results caused by the inability to accurately identify the search intent that matches the query term.
  • the user's search intent can be accurately identified, and the recall cluster can be determined according to the distance between the landmark materials in the case that other existing search strategies cannot recall the search results.
  • the landmark material at the center point improves the accuracy of the recall search results.
  • FIG. 2 is a schematic structural diagram of search logic of an example of the present application.
  • FIG. 3 is a schematic structural diagram of search logic of an example of the present application.
  • FIG. 4 is a schematic diagram showing the hardware structure of a search device of an example of the present application.
  • the search items corresponding to the search intent can be stored in the database.
  • the search intent corresponding to the query term may be determined according to the textual relevance of the query term and the searched material in the database.
  • strict text-related criteria may be set, resulting in the inability to recognize the search intent of some users and the inability to recall search results.
  • the application example discloses a search method. As shown in FIG. 1 , the method includes: Step 100 to Step 140.
  • Step 100 Match the obtained query words with the landmark materials in the preset landmark database.
  • the preset landmark database may include a plurality of first landmark materials, and each of the first landmark materials may include at least: a landmark name and a geographical location of the landmark.
  • the geographical location of the landmark is usually represented by the latitude and longitude coordinates of the landmark.
  • the landmark name may further include a corresponding abbreviation, full name, alias, Chinese character name, digital name, and the like.
  • the landmark name in the landmark database may include landmarks in the form of "Beijing No. 18 Middle School (18)", "Peking University", and the like.
  • the query word may be a query word manually input by the user through the input interface of the search platform, or may be a keyword corresponding to the link extracted by the page program after the user clicks the link on the search platform page, or the search channel of the user through the search platform. Select the keywords you enter, such as the business name, place name, and more. This application does not limit the way in which query terms are obtained.
  • the search engine After the query word is obtained, the search engine performs a matching operation in the preset landmark database according to the obtained query word.
  • the obtained query words are respectively fuzzyly matched with the name of each first landmark material in the preset landmark database (including a short name or full name, an alias, a Chinese character name, a digital name, etc.), and the text correlation is selected.
  • the first landmark material corresponding to the name of the landmark material satisfying the preset condition is used as the second landmark material matching the query term.
  • a plurality of second landmark materials that match the query term can be obtained.
  • Step 110 Cluster the second landmark materials that match the query words based on the physical distance between the geographic locations of the second landmark materials.
  • the obtained plurality of second landmark materials matching the query word are clustered, and the second landmark materials with similar distances are aggregated into one cluster, and multiple clusters are obtained.
  • a plurality of second landmark materials can be included in each cluster.
  • the clustering algorithm such as K-MEANS algorithm, K-MEDOIDS algorithm and CLARANS algorithm may be used to cluster the second landmark materials matching the query words according to the geographic location.
  • the specific method for clustering the second landmark material based on the distance between the second landmark materials is as follows: the second landmark material matching the query word is used as a cluster sample, and is preset.
  • the distance threshold D th between the landmark materials is used as a constraint, and the distance between any two cluster samples is iteratively calculated and determined until all cluster samples are aggregated to at least one cluster.
  • Input a second landmark material that matches the query term
  • Specific algorithm Calculate the distance between two samples in the cluster sample (the second landmark material), and take the minimum distance D min . If the minimum distance Dmin is within a preset distance threshold Dth , merge two samples corresponding to the minimum distance Dmin , such as samples A and B, that is, two samples A corresponding to the minimum distance Dmin B generates a cluster sample C and deletes two samples A and B corresponding to the minimum distance D min .
  • a cluster sample C (landmark material) based on two cluster samples A and B corresponding to the minimum distance D min , taking the latitude and longitude coordinates of the intermediate point of the geographical position of the two cluster samples as the generated cluster sample C Geographic location.
  • the above process of calculating distance and sample merging is repeated until all cluster samples are aggregated into one cluster, or the distance between the most recent two cluster samples is greater than a preset distance threshold D th .
  • each cluster corresponding to the cluster sample is obtained, wherein each cluster includes at least one geographic location, and the geographic location may be a geographic location of the second landmark material, or according to the second landmark
  • the geographic location of the clustered sample generated by the geographic location of the item.
  • the landmark material corresponding to each cluster can be determined.
  • the geographic location in each cluster obtained by the clustering may be traversed, and the second landmark material closest to the geographical location is taken as the landmark material in the preset landmark database corresponding to each cluster.
  • Step 120 Determine a correlation between the query term and the search intent according to the clustering result.
  • the degree of aggregation of clustering results reflects the relevance of query terms to landmark searches.
  • the ratio of the number of corresponding second landmark materials in the largest cluster obtained by clustering to the number of second landmark materials matching the query word indicates the degree of aggregation of the clustering results, as a query word and a search intention ( For example, the relevance of landmark search).
  • a correlation between the query term and the search intent is a correlation between the plurality of second landmark materials in the clustering result.
  • the relevance of the query term to the landmark search can be directly determined based on the clustering results. The more the number of corresponding landmark materials in the largest cluster obtained by clustering, the higher the degree of aggregation of landmark materials with good matching and the stronger the correlation.
  • each geographic location included in each cluster obtained by clustering corresponds to one landmark material. Determining the correlation between the query word and the landmark search according to the clustering result, comprising: determining a ratio of the number of landmark materials in the largest cluster obtained by the cluster and the number of second landmark materials matching the query word successfully; using the ratio as The relevance of query terms to landmark searches.
  • Step 130 If the correlation is greater than a preset threshold, determine that the user's search intent is a landmark search.
  • determining that the search intent is a landmark search.
  • the preset threshold may be a value less than 1, such as 70%. If the determined correlation is greater than the preset threshold, it is determined that the user's search intent is a landmark search; otherwise, the user's search intent is not a landmark search. For example, if the ratio of the number of landmark materials corresponding to the geographical location included in the largest cluster and the number of the second landmark materials matching the query word is greater than 70% in the clustering result, it is determined that the user's search intention is a landmark search. Otherwise, the user's search intent is not considered a landmark search.
  • the threshold can be set to a value between 60% and 90%. If the preset threshold is set to a lower value, indicating that the judgment condition of the clustering result is relaxed, the accuracy of the search is correspondingly reduced. If the preset threshold is set to a higher value, indicating the judgment condition of the strict clustering result, the accuracy of the search will be correspondingly increased, which may result in fewer matching search results.
  • the cluster is the largest cluster, that is, the geographic location (corresponding to the landmark material) included in the cluster containing the most geographical locations, the total number of geographical locations used for clustering (equivalent to the total number of cluster samples) The ratio is greater than 70%, and the landmarks are considered to be highly aggregated, and the user's search intent is determined to be a landmark search. Then, the landmark material closest to the center point of the largest cluster obtained by clustering can be taken as the landmark that is queried for the user.
  • Step 140 If the search intention of the user is a landmark search, determine a third landmark material according to the clustering result, and recall the third landmark material.
  • determining the third landmark material comprises: determining a geographic location center point of the largest cluster obtained by clustering; determining a landmark material closest to a geographic center point of the largest cluster as the third landmark material.
  • each cluster includes a plurality of landmark materials, each landmark material corresponds to a geographic location, and the geographic location is an original geographic location of the landmark material. Or a geographic location based on the geographic location of the landmark material.
  • the landmark material corresponding to each cluster can be determined. In a specific implementation, first determining a geographic location center point in the largest cluster obtained by clustering; then, traversing the cluster sample to determine a second landmark material in the cluster sample that is closest to the geographic center point, and the landmark material is As a third landmark material recall for the user.
  • the process of determining the geographic center point in the largest cluster obtained by clustering is a process of determining the center point of a plurality of geographic locations in the largest cluster.
  • the process of determining the cluster sample closest to the geographic center point is the process of calculating the distance between the geographic center point and the plurality of geographic locations, and determining the minimum distance.
  • the material search is performed using a default search strategy.
  • the search method disclosed in the embodiment of the present application matches the obtained query words with the landmark materials in the preset landmark database, and then clusters the landmark materials that are successfully matched based on the distance between the landmark materials, according to the clustering. As a result, the correlation between the query word and the landmark search is determined. If the correlation is greater than the preset threshold, determining that the user's search intention is a landmark search, and if the user's search intention is a landmark search, performing a material recall according to the clustering result. .
  • the user's search intention can be accurately identified, and the landmark material of the recall cluster center point is determined according to the distance between the landmark materials, and the accuracy of the recall search result is improved.
  • a search method is also disclosed in the example of the present application, in which the matching query words are matched with the first landmark material in the preset landmark database, including: based on Text relevance, the obtained query words are fuzzy matched with each first landmark material in the preset landmark database.
  • the first text relevance determination threshold and the second text correlation determination are preset. Threshold.
  • the first text relevance determination threshold is a text relevance determination threshold for determining whether the query word matches the landmark name;
  • the second text relevance determination threshold is determining the query word in the search strategy (eg, merchant strategy, landmark strategy)
  • the text relevance judgment threshold for whether the search item in the database matches. For example, the first text relevance determination threshold is lower than the second text relevance determination threshold.
  • a looser first text relevance determination threshold is set, such as setting the first text relevance determination threshold to a text relevance score higher than 80 points.
  • the threshold of the first text relevance judgment is set because the looseness is set. Therefore, the query is made.
  • the words “Ningmen Gate” and the “National People’s University” and “People’s University Ximen Barbecue Restaurant” in the pre-set landmark database can be successfully matched.
  • the query words can be pre-processed, such as the way of extracting the core words, and the query words are fuzzy matched with the landmark materials.
  • the query term “Ximen Ximen” as an example, the unimportant word “Ximen” can be discarded, and the core word “NPC” can be extracted to match the landmark materials in the preset landmark database. Then the landmark material “National People’s Dormitory Building No. 1” Can match successfully.
  • fuzzy matching of the obtained query words with each landmark material in the preset landmark database can ensure that the recalled landmark materials satisfy the text correlation and ensure a basic user experience.
  • FIG. 4 is a hardware structural diagram of a search provided by some embodiments of the present application.
  • the search device 40 can include a processor 41 and a non-transitory computer readable storage medium 42.
  • the processor 41 and the non-transitory computer readable storage medium 42 can communicate via the system bus 43.
  • the processor 41 can perform the search method described above.
  • the non-transitory computer readable storage medium 42 referred to herein can be any electronic, magnetic, optical, or other physical storage device that can contain or store information such as executable instructions, data, and the like.
  • the non-transitory computer readable storage medium 92 may be a RAM (Random Access Memory), a volatile memory, a nonvolatile memory, a flash memory, a storage drive (such as a hard disk drive), a solid state drive, Any type of storage disk (such as a compact disc, DVD, etc.), or a similar storage medium, or a combination thereof.
  • the search logic 20 can include:
  • the text matching module 200 is configured to match the obtained query word with the first landmark material in the preset landmark database, and determine a plurality of second landmark materials that match the query word;
  • the clustering module 210 is configured to cluster the plurality of second landmark materials based on a physical distance between respective geographic locations of the plurality of second landmark materials to obtain a clustering result;
  • the correlation determining module 220 is configured to determine, according to the clustering result, a correlation between the query word and a landmark search;
  • the material recalling module 240 is configured to determine a third landmark material according to the clustering result and recall the third landmark material if the correlation is greater than a preset threshold.
  • the search device further includes an intent identification module 230.
  • the intent identification module 230 is configured to determine that the user's search intent is a landmark search if the correlation is greater than a preset threshold.
  • the material recall module 240 is configured to determine a third landmark material according to the clustering result if the user's search intention is a landmark search, and recall the third landmark material.
  • the correlation determination module 220 includes:
  • the ratio determining unit 2201 is configured to determine a ratio of the quantity of the landmark material corresponding to the largest cluster in the clustering result to the quantity of the second landmark material;
  • the correlation determining unit 2202 is configured to use the ratio as a correlation between the query word and the landmark search.
  • the text matching module 200 is specifically configured to: perform fuzzy matching on the first landmark material in the landmark database based on text relevance.
  • the material recall module 240 includes:
  • a central point determining unit 2401 configured to determine a geographic center point of the largest cluster of the clustering results
  • the landmark material recalling unit 2402 is configured to use the landmark material closest to the geographic center point of the largest cluster as the third landmark material.
  • the second landmark material is used as a cluster sample when clustering the plurality of second landmark materials.
  • the clustering module 210 is configured to calculate a distance between each two cluster samples in the cluster sample to obtain a minimum distance; if the minimum distance is smaller than the distance threshold, according to the two corresponding to the minimum distance
  • the clustering sample generates a new clustering sample, and deletes two clustering samples corresponding to the minimum distance, wherein the geographical location corresponding to the newly generated clustering sample is an intermediate point of the geographic location of the two clustering samples;
  • the clustering is completed when the remaining cluster samples are aggregated into one cluster, or the distance between each two remaining cluster samples is greater than the distance threshold; otherwise, each of the two cluster samples is calculated again. The distance between clustered samples.
  • the search device disclosed in the embodiment matches the acquired query words with the landmark materials in the preset landmark database, and then clusters the landmark materials that are successfully matched based on the distance between the landmark materials, and obtains according to the obtained The clustering result determines the relevance of the landmark material that is successfully matched. Finally, if the correlation is greater than the preset threshold, the user's search intent is determined to be a landmark search, and the material recall is performed according to the clustering result, thereby solving the existing There is a problem in the technology that cannot accurately identify the search intent that matches the query term, and cannot recall the search result.
  • the user's search intent can be accurately identified, and the recall cluster can be determined according to the distance between the landmark materials in the case that other existing search strategies cannot recall the search results.
  • the landmark material at the center point improves the accuracy of the recall search results.
  • the present application also discloses an electronic device comprising a memory, a processor, and a computer program stored on the memory and operable on the processor, wherein the processor implements the computer program to implement the present application
  • the search method described in the first embodiment and the second embodiment can be a PC, a mobile terminal, a personal digital assistant, a tablet, or the like.
  • the present application also discloses a computer readable storage medium having stored thereon a computer program, the program being executed by the processor to implement the steps of the search method as described in the first embodiment and the second embodiment of the present application.
  • Embodiments of the subject matter and functional operations described in this specification can be implemented in the following: digital electronic circuits, tangible embodied computer software or firmware, computer hardware including the structures disclosed in the specification and their structural equivalents, or One or more combinations.
  • Embodiments of the subject matter described in this specification can be implemented as one or more computer programs, i.e., one of computer program instructions encoded on a tangible, non-transitory program carrier to be executed by a data processing device or to control operation of a data processing device or Multiple modules.
  • the program instructions may be encoded on an artificially generated propagating signal, such as a machine-generated electrical, optical or electromagnetic signal that is generated to encode and transmit the information to a suitable receiver device for data
  • the processing device executes.
  • the computer storage medium can be a machine readable storage device, a machine readable storage substrate, a random or serial access memory device, or a combination of one or more of them.
  • the processes and logic flows described in this specification can be performed by one or more programmable computers executing one or more computer programs to perform the corresponding functions by operating in accordance with input data and generating an output.
  • the processing and logic flow may also be performed by dedicated logic circuitry, such as an FPGA (Field Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit), and the apparatus may also be implemented as dedicated logic circuitry.
  • FPGA Field Programmable Gate Array
  • ASIC Application Specific Integrated Circuit
  • Computers suitable for the execution of a computer program include, for example, a general purpose and/or special purpose microprocessor, or any other type of central processing unit.
  • the central processing unit will receive instructions and data from a read only memory and/or a random access memory.
  • the basic components of a computer include a central processing unit for implementing or executing instructions and one or more memory devices for storing instructions and data.
  • the computer will also include one or more mass storage devices for storing data, such as a magnetic disk, magneto-optical disk or optical disk, or the like, or the computer will be operatively coupled to the mass storage device for receiving data or It transmits data, or both.
  • the computer does not have to have such a device.
  • the computer can be embedded in another device, such as a mobile phone, a personal digital assistant (PDA), a mobile audio or video player, a game console, a global positioning system (GPS) receiver, or, for example, a universal serial bus (USB) ) Portable storage devices for flash drives, to name a few.
  • PDA personal digital assistant
  • GPS global positioning system
  • USB universal serial bus
  • Computer readable media suitable for storing computer program instructions and data include all forms of nonvolatile memory, media and memory devices including, for example, semiconductor memory devices (eg, EPROM, EEPROM, and flash memory devices), magnetic disks (eg, internal hard drives or Mobile disk), magneto-optical disks, and CD ROM and DVD-ROM disks.
  • semiconductor memory devices eg, EPROM, EEPROM, and flash memory devices
  • magnetic disks eg, internal hard drives or Mobile disk
  • magneto-optical disks e.g, CD ROM and DVD-ROM disks.
  • the processor and memory may be supplemented by, or incorporated in, special purpose logic circuitry.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

一种搜索方法、搜索装置和计算机可读存储介质。在一示例中,将获取的查询词与预设地标数据库中的第一地标物料进行匹配,确定与所述查询词相匹配的多个第二地标物料;基于所述多个第二地标物料各自对应的地理位置之间的物理距离,对所述多个第二地标物料进行聚类,得到聚类结果;根据所述聚类结果,确定所述查询词与搜索意图之间的相关性;根据所述搜索意图,召回相匹配的第三地标物料。

Description

搜索方法、装置及非临时性计算机可读存储介质
相关申请的交叉引用
本专利申请要求于2017年1月20日提交的、申请号为201710042949.1、发明名称为“一种搜索方法及装置、电子设备”的中国专利申请的优先权,该申请的全文以引用的方式并入本文中。
技术领域
本申请涉及搜索技术领域,特别是涉及搜索方法、装置及非临时性计算机可读存储介质。
背景技术
在搜索技术领域,可获取到查询词,搜索引擎会首先根据查询词确定用户的搜索意图,然后,根据用户的搜索意图选择相应的搜素策略执行搜索操作。
发明内容
本申请提供一种搜索方法,解决现有技术中存在的无法识别与查询词匹配的搜索意图而导致的搜索结果不准确的问题。
为了解决上述问题,第一方面,本申请实施例提供了一种搜索方法,包括:
将获取的查询词与预设地标数据库中的第一地标物料进行匹配,确定与所述查询词相匹配的多个第二地标物料;
基于所述多个第二地标物料各自对应的地理位置之间的物理距离,对所述多个第二地标物料进行聚类,得到聚类结果;
根据所述聚类结果,确定所述查询词与搜索意图之间的相关性;
根据所述搜索意图,召回相匹配的第三地标物料。
第二方面,本申请实施例提供了一种搜索装置,包括:
处理器和非临时性计算机可读存储介质,
所述非临时性计算机可读存储介质存储有能够被所述处理器执行的机器可执行指令,所述处理器被所述机器可执行指令促使:
将获取的查询词与预设地标数据库中的第一地标物料进行匹配,确定与所述查询词相匹配的多个第二地标物料;
基于所述多个第二地标物料各自对应的地理位置之间的物理距离,对所述多个第二地标物料进行聚类,得到聚类结果;
根据所述聚类结果,确定所述查询词与搜索意图之间的相关性;
根据所述搜索意图,召回相匹配的第三地标物料。
第三方面,本申请实施例提供了一种非临时性计算机可读存储介质,存储有机器可执行指令,在被网络设备的处理器调用和执行时,所述机器可执行指令促使所述处理器执行:
将获取的查询词与预设地标数据库中的第一地标物料进行匹配,确定与所述查询词相匹配的多个第二地标物料;
基于所述多个第二地标物料各自对应的地理位置之间的物理距离,对所述多个第二地标物料进行聚类,得到聚类结果;
根据所述聚类结果,确定所述查询词与搜索意图之间的相关性;
根据所述搜索意图,召回相匹配的第三地标物料。
本申请实施例公开的搜索方法,通过将获取的查询词与预设地标数据库中的地标物料进行匹配,然后基于地标物料之间的距离对匹配成功的所述地标物料进行聚类,根据聚类结果确定匹配成功的所述地标物料的相关性,若所述相关性大于预设阈值,则确定用户的搜索意图为地标搜索,并进一步根据聚类结果执行物料召回,解决了现有技术中存在的无法准确识别与查询词匹配的搜索意图的问题从而导致的搜索结果不准确的问题。通过结合文本匹配与聚类方法,确定用户的搜索意图,可以准确地识别用户搜索意图,并在采用其他既有搜索策略无法召回搜索结果的情况下,根据地标物料之间的距离确定召回聚类中心点的地标物料,提高了召回搜索结果的准确性。
附图说明
为了更清楚地说明本申请实施例的技术方案,下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本申请的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动性的前提下,还可以根据这些附图获得其他的附图。
图1是本申请示例的搜索方法的流程图;
图2是本申请示例的搜索逻辑的结构示意图;
图3是本申请示例的搜索逻辑的结构示意图;
图4是本申请示例的搜索装置的硬件结构示意图。
具体实施方式
下面将结合本申请实施例中的附图,对本申请实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例仅仅是本申请一部分实施例,而不是全部的实施例。基于本申请中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本申请保护的范围。
可将搜索意图对应的搜索物料存储在数据库中。在搜索过程中,可根据查询词与数据库中的搜索物料的文本相关性确定该查询词对应的搜索意图。但,为了保证识别的准确性,可能会设置严格的文本相关准则,导致无法识别某些用户的搜索意图,无法召回搜索结果。
本申请示例公开了一种搜索方法,如图1所示,该方法包括:步骤100至步骤140。
步骤100,将获取的查询词与预设地标数据库中的地标物料进行匹配。
具体实施时,预设地标数据库中可包括多条第一地标物料,每条第一地标物料可至少包括:地标名称、地标的地理位置。其中,地标的地理位置通常由地标的经纬度坐标表示。为了提高文本匹配的鲁棒性,地标名称还可包括对应有相应的简称、全称、别名、汉字名称、数字名称等。例如,地标数据库中的地标名称可包括“北京市第十八中学(18中)”、“北京大学(Peking University)”等形式的地标。
查询词可以为用户通过搜索平台的输入界面手动输入的查询词,也可以是用户点击搜索平台页面上的链接后由页面程序提取的该链接对应的关键词,或者是用户通过搜索平台的搜索频道选择输入的关键词,例如,商家名称、地标名称等。本申请对获取查询词的方式不做 限定。
在获取到查询词之后,搜索引擎会根据获取的查询词在预设地标数据库中执行匹配操作。例如,将获取的所述查询词分别与所述预设地标数据库中每条第一地标物料的名称(可包括简称或全称、别名、汉字名称、数字名称等)进行模糊匹配,选择文本相关性满足预设条件的地标物料名称对应的第一地标物料作为与该查询词相匹配的第二地标物料。在一示例中,经过模糊匹配,将可获取到与所述查询词相匹配的多条第二地标物料。
步骤110,基于第二地标物料地理位置之间的物理距离,对与查询词相匹配的所述第二地标物料进行聚类。
根据第二地标物料的地理位置,对获取到的与所述查询词匹配的多条第二地标物料进行聚类,将距离相近的第二地标物料聚集到一个簇中,可以得到多个簇。每个簇中可包括多个第二地标物料。具体实施时,可以采用:K-MEANS算法、K-MEDOIDS算法、CLARANS算法等聚类算法对与查询词匹配的第二地标物料按照地理位置进行聚类。
在一示例中,基于第二地标物料之间的距离对所述第二地标物料进行聚类的具体方法如下:以与所述查询词匹配的第二地标物料作为聚类样本,以预设的地标物料之间的距离阈值D th作为约束,迭代计算并判断任意两个聚类样本之间的距离,直至所有聚类样本聚集到至少一个簇。
输入:与所述查询词匹配的第二地标物料;
特征:两个第二地标物料之间的距离;
具体算法:计算聚类样本(第二地标物料)中两两样本之间的距离,取其中的最小距离D min。如果所述最小距离D min在预设的距离阈值D th范围内,合并所述最小距离D min对应的两个样本,如样本A和B,即根据最小距离D min对应的两个样本A和B生成一个聚类样本C,并删除最小距离D min对应的两个样本A和B。在根据最小距离D min对应的两个聚类样本A和B生成一个聚类样本C(地标物料)时,取两个聚类样本的地理位置的中间点的经纬度坐标作为生成的聚类样本C的地理位置。
重复上述计算距离和样本合并的过程,直到所有的聚类样本都聚集到一个簇,或者最近的两个聚类样本之间的距离大于预设的距离阈值D th
经过前述聚类过程,将得到所述聚类样本对应的至少一个簇,其中,每个簇中包括至少一个地理位置,所述地理位置可为第二地标物料的地理位置,或者根据第二地标物料的地理位置生成的聚类样本的地理位置。根据聚类得到的每个簇中的地理位置,可以确定每个簇对 应的地标物料。具体实施时,可以遍历聚类得到的每个簇中的地理位置,取与该地理位置最近的第二地标物料,作为每个簇对应的预设地标数据库中的地标物料。
步骤120,根据聚类结果确定该查询词与搜索意图的相关性。
聚类结果的聚集程度反映了查询词与地标搜索的相关性。具体实施时,通过聚类得到的最大簇中对应的第二地标物料的数量和与查询词相匹配的第二地标物料的数量的比例表示聚类结果的聚集程度,作为查询词与搜索意图(例如,地标搜索)的相关性。所述查询词与所述搜索意图之间的相关性为所述聚类结果中所述多个第二地标物料之间的相关性。在一示例中,可根据聚类结果直接确定该查询词与地标搜索的相关性。聚类得到的最大簇中对应的地标物料的数量越多,说明匹配成功的地标物料的聚集程度越高,相关性越强。
具体实施时,聚类得到的每个簇中包含的每一个地理位置对应一个地标物料。根据聚类结果,确定查询词与地标搜索的相关性,包括:确定聚类得到的最大簇中地标物料的数量和与查询词匹配成功的第二地标物料的数量的比值;将所述比值作为查询词与地标搜索的相关性。
步骤130,若所述相关性大于预设阈值,则确定用户的搜索意图为地标搜索。
在一示例中,若所述聚类结果中匹配成功的多个第二地标物料之间的相关性大于预设阈值,确定所述搜索意图为地标搜索。
具体实施时,所述预设阈值可以为小于1的数值,如70%。如果确定的相关性大于预设阈值,则确定用户的搜索意图为地标搜索,否则,认为用户的搜索意图不是地标搜索。例如,如果聚类结果中,最大簇中包含的地理位置对应的地标物料的数量和与查询词相匹配的第二地标物料的数量的比例大于70%,则确定用户的搜索意图为地标搜索,否则,认为用户的搜索意图不是地标搜索。
在一示例中,该阈值可以设置为60%至90%之间的数值。如果预设阈值设置为较低的数值,表示放宽了聚类结果的判断条件,则搜索的准确率会相应降低。如果预设阈值设置为较高的数值,表示严格聚类结果的判断条件,则搜索的准确率会相应提高,则可能导致匹配的搜索结果较少。
具体实施时,如果聚类得到的最大簇,即包含最多地理位置的簇所包含的地理位置(对应于地标物料)的个数占用于聚类的总地理位置的总数(相当于聚类样本总数)的比例大于70%,认为地标的聚集性很高,确定用户的搜索意图为地标搜索。然后,可以取与聚类得到的最大簇的中心点距离最近的地标物料,作为为用户查询到的地标。
步骤140,若所述用户的搜索意图为地标搜索,根据聚类结果确定第三地标物料,并召回该第三地标物料。
具体实施时,确定第三地标物料,包括:确定聚类得到的最大簇的地理位置中心点;将距离所述最大簇的地理位置中心点最近的地标物料确定为第三地标物料。
经过前述聚类过程,将得到所述聚类样本对应的多个簇,每个簇中包括多个地标物料,每个地标物料对应一个地理位置,所述地理位置为地标物料的原始地理位置,或者根据地标物料的地理位置生成的地理位置。根据聚类得到的每个簇中的地理位置,可以确定每个簇对应的地标物料。具体实施时,首先确定聚类得到的最大簇中的地理位置中心点;然后,遍历聚类样本,确定聚类样本中与所述地理位置中心点距离最近的第二地标物料,将该地标物料作为为用户查询到的第三地标物料召回。确定聚类得到的最大簇中的地理位置中心点的过程,是确定该最大簇中多个地理位置的中心点的过程。确定与所述地理位置中心点距离最近的聚类样本的过程,即是计算地理位置中心点分别与多个地理位置之间的距离,并确定最小距离的过程。
若确定所述用户的搜索意图非地标搜索,则采用默认的搜索策略执行物料召回。
本申请实施例公开的搜索方法,通过将获取的查询词与预设地标数据库中的地标物料进行匹配,然后基于地标物料之间的距离对匹配成功的所述地标物料进行聚类,根据聚类结果确定查询词与地标搜索的相关性,若所述相关性大于预设阈值,则确定用户的搜索意图为地标搜索,若所述用户的搜索意图为地标搜索,则根据聚类结果执行物料召回。通过结合文本匹配与聚类方法,确定用户的搜索意图,可以准确地识别用户搜索意图,并根据地标物料之间的距离确定召回聚类中心点的地标物料,提高了召回搜索结果的准确性。
基于如图1所述的方法,在本申请示例还公开的一种搜索方法,在该方法中,所述将获取的查询词与预设地标数据库中的第一地标物料进行匹配,包括:基于文本相关性,将获取的查询词与预设地标数据库中每个第一地标物料进行模糊匹配。
具体实施时,在将获取的查询词与所述预设地标数据库中每条第一地标物料的名称基于文本相关性进行匹配时,预先设置第一文本相关性判断阈值和第二文本相关性判断阈值。其中,第一文本相关性判断阈值为判断查询词与地标名称是否匹配的文本相关性判断阈值;第二文本相关性判断阈值为在搜索策略(例如,商家策略、地标策略)中判断查询词与数据库中的搜索物料是否匹配的文本相关性判断阈值。例如,第一文本相关性判断阈值低于第二文本相关性判断阈值。
以查询词为“人大西门”为例,假设既有搜索物料中包含“人大”的搜索物料,并设置一个文本相关性判断阈值并根据搜索策略(例如,商家策略、地标策略等)判断“人大西门”和“人大”两个词的文本相关性。由于该文本相关性判断阈值设置较严格,如将文本相关性判断阈值设置为文本相关性得分高于90分,因此,导致查询词“人大西门”无法与搜索物料“人大”匹配成功。
本实施例中,设置了较宽松的第一文本相关性判断阈值,如将第一文本相关性判断阈值设置为文本相关性得分高于80分。当用查询词“人大西门”与预设地标数据库中的“人大”、“人民大学西门烧烤店”等地标物料进行匹配时,由于设置了较宽松的第一文本相关性判断阈值,因此,查询词“人大西门”与预设地标数据库中的“人大”、“人民大学西门烧烤店”等地标物料均可以匹配成功。
具体实施时,还可以通过对查询词进行预处理,如提取核心词的方式,将查询词与地标物料进行模糊匹配。以查询词为“人大西门”为例,可以丢弃不重要的词“西门”,提取核心词“人大”与预设地标数据库中的地标物料进行匹配,那么地标物料“人大宿舍一号楼”也可以匹配成功。
基于文本相关性,将获取的查询词与预设地标数据库中每个地标物料进行模糊匹配能够保证召回的地标物料满足文本相关,保证基本的用户体验。
以上对本申请提供的方法进行了描述。下面对本申请提供的装置进行描述:
参见图4,图4为本申请一些实施例提供的搜索的硬件结构图。该搜索装置40可包括处理器41以及非临时性计算机可读存储介质42。其中,处理器41和非临时性计算机可读存储介质42可经由系统总线43通信。并且,通过读取并执行非临时性计算机可读存储介质42中存储的与搜索逻辑20对应的机器可执行指令,处理器41可执行上文所述的搜索方法。
本文提到的非临时性计算机可读存储介质42可以是任何电子、磁性、光学或其他物理存储装置,可以包含或存储信息,如可执行指令、数据,等等。例如,所述非临时性计算机可读存储介质92可以是RAM(Random Access Memory,随机存取存储器)、易失存储器、非易失性存储器、闪存、存储驱动器(如硬盘驱动器)、固态硬盘、任何类型的存储盘(如光盘、DVD等),或者类似的存储介质,或者它们的组合。
如图2所示,从功能上划分,所述搜索逻辑20可包括:
文本匹配模块200,用于将获取的查询词与预设地标数据库中的第一地标物料进行匹配,确定与所述查询词相匹配的多个第二地标物料;
聚类模块210,用于基于所述多个第二地标物料各自对应的地理位置之间的物理距离,对所述多个第二地标物料进行聚类,得到聚类结果;
相关性确定模块220,用于根据所述聚类结果,确定所述查询词与地标搜索之间的相关性;
物料召回模块240,用于若所述相关性大于预设阈值,根据所述聚类结果确定第三地标物料,并召回所述第三地标物料。
在一示例中,所述搜索装置进一步包括意图识别模块230。
所述意图识别模块230用于若所述相关性大于预设阈值,则确定用户的搜索意图为地标搜索。
所述物料召回模块240用于若用户的搜索意图为地标搜索,根据所述聚类结果确定第三地标物料,并召回所述第三地标物料。
在一示例中,如图3所示,所述相关性确定模块220包括:
比例确定单元2201,用于确定所述聚类结果中的最大簇对应的地标物料的数量与所述第二地标物料的数量的比值;
相关性确定单元2202,用于将所述比值作为所述查询词与地标搜索之间的相关性。
可选的,所述文本匹配模块200具体用于:基于文本相关性,将所述查询词与所述地标数据库中每个第一地标物料进行模糊匹配。
在一示例中,如图3所示,所述物料召回模块240包括:
中心点确定单元2401,用于确定所述聚类结果中的最大簇的地理位置中心点;
地标物料召回单元2402,用于将距离所述最大簇的地理位置中心点最近的地标物料作为所述第三地标物料。
在一示例中,在对所述多个第二地标物料进行聚类时,所述第二地标物料作为聚类样本。
所述聚类模块210用于计算所述聚类样本中每两个聚类样本之间的距离,得到最小距离;如果所述最小距离小于所述距离阈值,根据该在最小距离对应的两个聚类样本生成新的聚类样本,并删除该最小距离对应的两个聚类样本,其中,该新生成的聚类样本对应的地理位置为所述两个聚类样本地理位置的中间点;当剩余的聚类样本聚集到一个簇,或者每两个剩余的聚类样本之间的距离大于所述距离阈值,所述聚类完成;否则,再次执行计算所述聚类样本中每两个聚类样本之间的距离。
本实施例中公开的搜索装置的各模块的具体实施方式,参见实施例一和实施例二的相关部分,此处不再赘述。
本实施例中公开的搜索装置,通过将获取的查询词与预设地标数据库中的地标物料进行匹配,然后基于地标物料之间的距离对匹配成功的所述地标物料进行聚类,并根据获得的聚类结果确定匹配成功的所述地标物料的相关性,最后若所述相关性大于预设阈值,则确定用户的搜索意图为地标搜索,并根据聚类结果执行物料召回,解决了现有技术中存在的无法准确识别与查询词匹配的搜索意图,无法召回搜索结果的问题。通过结合文本匹配与聚类方法,确定用户的搜索意图,可以准确地识别用户搜索意图,并在采用其他既有搜索策略无法召回搜索结果的情况下,根据地标物料之间的距离确定召回聚类中心点的地标物料,提高了召回搜索结果的准确性。
本申请还公开了一种电子设备,包括存储器、处理器及存储在所述存储器上并可在处理器上运行的计算机程序,其特征在于,所述处理器执行所述计算机程序时实现本申请实施例一和实施例二所述的搜索方法。所述电子设备可以为PC机、移动终端、个人数字助理、平板电脑等。
本申请还公开了一种计算机可读存储介质,其上存储有计算机程序,该程序被处理器执行时实现如本申请实施例一和实施例二所述的搜索方法的步骤。
本说明书中的各个实施例均采用递进的方式描述,每个实施例重点说明的都是与其他实施例的不同之处,各个实施例之间相同相似的部分互相参见即可。对于装置实施例而言,由于其与方法实施例基本相似,所以描述的比较简单,相关之处参见方法实施例的部分说明即可。
以上对本申请提供的一种搜索方法及装置进行了详细介绍,本文中应用了具体个例对本申请的原理及实施方式进行了阐述,以上实施例的说明只是用于帮助理解本申请的方法及其核心思想;同时,对于本领域的一般技术人员,依据本申请的思想,在具体实施方式及应用范围上均会有改变之处,综上所述,本说明书内容不应理解为对本申请的限制。
通过以上的实施方式的描述,本领域的技术人员可以清楚地了解到各实施方式可借助软件加必需的通用硬件平台的方式来实现,当然也可以通过硬件实现。基于这样的理解,上述技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来,该计算机软件产品可以存储在计算机可读存储介质中,如ROM/RAM、磁碟、光盘等,包括若干指令用以使得一台计算机设备(可以是个人计算机,服务器,或者网络设备等)执行各个实施 例或者实施例的某些部分所述的方法。
本说明书中描述的主题及功能操作的实施例可以在以下中实现:数字电子电路、有形体现的计算机软件或固件、包括本说明书中公开的结构及其结构性等同物的计算机硬件、或者它们中的一个或多个的组合。本说明书中描述的主题的实施例可以实现为一个或多个计算机程序,即编码在有形非暂时性程序载体上以被数据处理装置执行或控制数据处理装置的操作的计算机程序指令中的一个或多个模块。可替代地或附加地,程序指令可以被编码在人工生成的传播信号上,例如机器生成的电、光或电磁信号,该信号被生成以将信息编码并传输到合适的接收机装置以由数据处理装置执行。计算机存储介质可以是机器可读存储设备、机器可读存储基板、随机或串行存取存储器设备、或它们中的一个或多个的组合。
本说明书中描述的处理及逻辑流程可以由执行一个或多个计算机程序的一个或多个可编程计算机执行,以通过根据输入数据进行操作并生成输出来执行相应的功能。所述处理及逻辑流程还可以由专用逻辑电路—例如FPGA(现场可编程门阵列)或ASIC(专用集成电路)来执行,并且装置也可以实现为专用逻辑电路。
适合用于执行计算机程序的计算机包括,例如通用和/或专用微处理器,或任何其他类型的中央处理单元。通常,中央处理单元将从只读存储器和/或随机存取存储器接收指令和数据。计算机的基本组件包括用于实施或执行指令的中央处理单元以及用于存储指令和数据的一个或多个存储器设备。通常,计算机还将包括用于存储数据的一个或多个大容量存储设备,例如磁盘、磁光盘或光盘等,或者计算机将可操作地与此大容量存储设备耦接以从其接收数据或向其传送数据,抑或两种情况兼而有之。然而,计算机不是必须具有这样的设备。此外,计算机可以嵌入在另一设备中,例如移动电话、个人数字助理(PDA)、移动音频或视频播放器、游戏操纵台、全球定位系统(GPS)接收机、或例如通用串行总线(USB)闪存驱动器的便携式存储设备,仅举几例。
适合于存储计算机程序指令和数据的计算机可读介质包括所有形式的非易失性存储器、媒介和存储器设备,例如包括半导体存储器设备(例如EPROM、EEPROM和闪存设备)、磁盘(例如内部硬盘或可移动盘)、磁光盘以及CD ROM和DVD-ROM盘。处理器和存储器可由专用逻辑电路补充或并入专用逻辑电路中。
虽然本说明书包含许多具体实施细节,但是这些不应被解释为限制任何发明的范围或所要求保护的范围,而是主要用于描述特定发明的具体实施例的特征。本说明书内在多个实施例中描述的某些特征也可以在单个实施例中被组合实施。另一方面,在单个实施例中描述的各种特征也可以在多个实施例中分开实施或以任何合适的子组合来实施。此外,虽然特征可 以如上所述在某些组合中起作用并且甚至最初如此要求保护,但是来自所要求保护的组合中的一个或多个特征在一些情况下可以从该组合中去除,并且所要求保护的组合可以指向子组合或子组合的变型。
类似地,虽然在附图中以特定顺序描绘了操作,但是这不应被理解为要求这些操作以所示的特定顺序执行或顺次执行、或者要求所有例示的操作被执行,以实现期望的结果。在某些情况下,多任务和并行处理可能是有利的。此外,上述实施例中的各种系统模块和组件的分离不应被理解为在所有实施例中均需要这样的分离,并且应当理解,所描述的程序组件和系统通常可以一起集成在单个软件产品中,或者封装成多个软件产品。
由此,主题的特定实施例已被描述。其他实施例在所附权利要求书的范围以内。在某些情况下,权利要求书中记载的动作可以以不同的顺序执行并且仍实现期望的结果。此外,附图中描绘的处理并非必需所示的特定顺序或顺次顺序,以实现期望的结果。在某些实现中,多任务和并行处理可能是有利的。
需要说明的是,在本文中,诸如第一和第二等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来,而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含,从而使得包括一系列要素的过程、方法、物品或者设备不仅包括那些要素,而且还包括没有明确列出的其他要素,或者是还包括为这种过程、方法、物品或者设备所固有的要素。在没有更多限制的情况下,由语句“包括一个……”限定的要素,并不排除在包括所述要素的过程、方法、物品或者设备中还存在另外的相同要素。
以上对本发明实施例所提供的方法和装置进行了详细介绍,本文中应用了具体个例对本发明的原理及实施方式进行了阐述,以上实施例的说明只是用于帮助理解本发明的方法及其核心思想;同时,对于本领域的一般技术人员,依据本发明的思想,在具体实施方式及应用范围上均会有改变之处,综上所述,本说明书内容不应理解为对本发明的限制。

Claims (15)

  1. 一种搜索方法,包括:
    将获取的查询词与预设地标数据库中的第一地标物料进行匹配,确定与所述查询词相匹配的多个第二地标物料;
    基于所述多个第二地标物料各自对应的地理位置之间的物理距离,对所述多个第二地标物料进行聚类,得到聚类结果;
    根据所述聚类结果,确定所述查询词与搜索意图之间的相关性;
    根据所述搜索意图,召回相匹配的第三地标物料。
  2. 根据权利要求1所述的方法,其中,所述查询词与所述搜索意图之间的相关性为所述聚类结果中所述多个第二地标物料之间的相关性,
    所述方法进一步包括:
    若所述聚类结果中匹配成功的多个第二地标物料之间的相关性大于预设阈值,确定所述搜索意图为地标搜索;
    所述根据所述搜索意图召回相匹配的地标物料,包括:
    若所述搜索意图为所述地标搜索,根据所述聚类结果确定所述第三地标物料,并召回所述第三地标物料。
  3. 根据权利要求1所述的方法,其中,根据所述聚类结果确定所述查询词与搜索意图之间的相关性,包括:
    确定所述聚类结果中的最大簇对应的地标物料的数量与所述第二地标物料的数量的比值;
    将所述比值作为所述查询词与地标搜索之间的相关性。
  4. 根据权利要求1所述的方法,其中,所述将获取的查询词与预设地标数据库中的第一地标物料进行匹配的步骤,包括:
    基于文本相关性,将所述查询词与所述地标数据库中每个第一地标物料进行模糊匹配。
  5. 根据权利要求2所述的方法,其中,根据所述聚类结果确定所述第三地标物料,包括:
    确定所述聚类结果中的最大簇的地理位置中心点;
    将距离所述最大簇的地理位置中心点最近的地标物料作为所述第三地标物料。
  6. 根据权利要求1所述的方法,其中,在对所述多个第二地标物料进行聚类时,所述第二地标物料作为聚类样本;
    对所述多个第二地标物料进行聚类,包括:
    计算所述聚类样本中每两个聚类样本之间的距离,得到最小距离;
    如果所述最小距离小于所述距离阈值,根据该在最小距离对应的两个聚类样本生成新的聚类样本,并删除该最小距离对应的两个聚类样本,其中,该新生成的聚类样本对应的地理位置为所述两个聚类样本地理位置的中间点;
    当剩余的聚类样本聚集到一个簇,或者每两个剩余的聚类样本之间的距离大于所述距离阈值,所述聚类完成;
    否则,再次执行计算所述聚类样本中每两个聚类样本之间的距离。
  7. 一种搜索装置,包括:
    处理器和非临时性计算机非临时性计算机可读存储介质,
    所述非临时性计算机非临时性计算机可读存储介质存储有能够被所述处理器执行的机器可执行指令,所述处理器被所述机器可执行指令促使:
    将获取的查询词与预设地标数据库中的第一地标物料进行匹配,确定与所述查询词相匹配的多个第二地标物料;
    基于所述多个第二地标物料各自对应的地理位置之间的物理距离,对所述多个第二地标物料进行聚类,得到聚类结果;
    根据所述聚类结果,确定所述查询词与搜索意图之间的相关性;
    根据所述搜索意图,召回相匹配的第三地标物料。
  8. 根据权利要求7所述的装置,其中,所述查询词与所述搜索意图之间的相关性为所述聚类结果中所述多个第二地标物料之间的相关性,
    所述处理器进一步被所述机器可执行指令促使:
    若所述聚类结果中匹配成功的多个第二地标物料之间的相关性大于预设阈值,确定所述搜索意图为地标搜索;
    所述处理器被所述机器可执行指令促使:
    若所述搜索意图为所述地标搜索,根据所述聚类结果确定第三地标物料,并召回所述第三地标物料。
  9. 根据权利要求7所述的装置,其中,所述处理器被所述机器可执行指令促使:
    确定所述聚类结果中的最大簇对应的地标物料的数量与所述第二地标物料的数量的比值;
    将所述比值作为所述查询词与地标搜索之间的相关性。
  10. 根据权利要求7所述的装置,其中,所述处理器被所述机器可执行指令促使:
    基于文本相关性,将所述查询词与所述地标数据库中每个第一地标物料进行模糊匹配。
  11. 根据权利要求8所述的装置,其中,所述处理器被所述机器可执行指令促使:
    确定所述聚类结果中的最大簇的地理位置中心点;
    将距离所述最大簇的地理位置中心点最近的地标物料作为所述第三地标物料。
  12. 根据权利要求7所述的装置,其中,在对所述多个第二地标物料进行聚类时,所述第二地标物料作为聚类样本;
    所述处理器被所述机器可执行指令促使:
    计算所述聚类样本中每两个聚类样本之间的距离,得到最小距离;
    如果所述最小距离小于所述距离阈值,根据该在最小距离对应的两个聚类样本生成新的聚类样本,并删除该最小距离对应的两个聚类样本,其中,该新生成的聚类样本对应的地理位置为所述两个聚类样本地理位置的中间点;
    当剩余的聚类样本聚集到一个簇,或者每两个剩余的聚类样本之间的距离大于所述距离阈值,所述聚类完成;
    否则,再次执行计算所述聚类样本中每两个聚类样本之间的距离。
  13. 一种非临时性计算机可读存储介质,存储有机器可执行指令,在被网络设备的处理器调用和执行时,所述机器可执行指令促使所述处理器执行:
    将获取的查询词与预设地标数据库中的第一地标物料进行匹配,确定与所述查询词相匹配的多个第二地标物料;
    基于所述多个第二地标物料各自对应的地理位置之间的物理距离,对所述多个第二地标物料进行聚类,得到聚类结果;
    根据所述聚类结果,确定所述查询词与搜索意图之间的相关性;
    根据所述搜索意图,召回相匹配的第三地标物料。
  14. 根据权利要求13所述的非临时性计算机可读存储介质,所述查询词与所述搜索意图之间的相关性为所述聚类结果中所述多个第二地标物料之间的相关性,
    其中,所述机器可执行指令进一步促使所述处理器执行:
    若所述聚类结果中匹配成功的多个第二地标物料之间的相关性大于预设阈值,确定所述搜索意图为地标搜索;
    所述机器可执行指令促使所述处理器执行:
    若所述搜索意图为所述地标搜索,根据所述聚类结果确定第三地标物料,并召回所述第三地标物料。
  15. 根据权利要求13所述的非临时性计算机非临时性计算机可读存储介质,其中,所述机器可执行指令促使所述处理器执行:
    确定所述聚类结果中的最大簇对应的地标物料的数量与所述第二地标物料的数量的比值;
    将所述比值作为所述查询词与地标搜索之间的相关性。
PCT/CN2017/119820 2017-01-20 2017-12-29 搜索方法、装置及非临时性计算机可读存储介质 Ceased WO2018133648A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CA3078148A CA3078148C (en) 2017-01-20 2017-12-29 SEARCH METHOD AND APPARATUS, AND NON-TEMPORARY COMPUTER-READABLE STORAGE MEDIA

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710042949.1 2017-01-20
CN201710042949.1A CN106933947B (zh) 2017-01-20 2017-01-20 一种搜索方法及装置、电子设备

Publications (1)

Publication Number Publication Date
WO2018133648A1 true WO2018133648A1 (zh) 2018-07-26

Family

ID=59424302

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/119820 Ceased WO2018133648A1 (zh) 2017-01-20 2017-12-29 搜索方法、装置及非临时性计算机可读存储介质

Country Status (4)

Country Link
CN (1) CN106933947B (zh)
CA (1) CA3078148C (zh)
TW (1) TWI669619B (zh)
WO (1) WO2018133648A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110874385A (zh) * 2018-08-10 2020-03-10 阿里巴巴集团控股有限公司 数据处理方法、装置和系统

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106933947B (zh) * 2017-01-20 2018-12-04 北京三快在线科技有限公司 一种搜索方法及装置、电子设备
CN109255023B (zh) * 2017-07-11 2021-09-14 中国移动通信集团浙江有限公司 线索信息处理方法及装置
CN108228820A (zh) * 2017-12-30 2018-06-29 厦门太迪智能科技有限公司 用户查询意图理解方法、系统及计算机终端
CN110362813B (zh) * 2018-04-09 2023-12-05 乐万家财富(北京)科技有限公司 基于bm25的搜索相关性度量方法、存储介质、设备及系统
CN108763538B (zh) * 2018-05-31 2019-07-23 北京嘀嘀无限科技发展有限公司 一种确定兴趣点poi地理位置的方法及装置
CN110674367B (zh) * 2019-09-09 2022-02-01 广州易起行信息技术有限公司 一种基于旅游行业产品的单汉字检索方法及装置
CN112989153B (zh) * 2019-12-13 2024-05-24 阿里巴巴集团控股有限公司 数据处理方法、装置及计算机设备
CN111400618B (zh) * 2020-02-14 2023-05-26 口口相传(北京)网络技术有限公司 一种数据搜索方法以及装置
CN113536156B (zh) * 2020-04-13 2024-05-28 百度在线网络技术(北京)有限公司 搜索结果排序方法、模型构建方法、装置、设备和介质
CN113779050B (zh) * 2020-06-23 2024-10-18 北京沃东天骏信息技术有限公司 一种客服机器人知识库的管理方法和装置

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103902694A (zh) * 2014-03-28 2014-07-02 哈尔滨工程大学 基于聚类和查询行为的检索结果排序方法
CN104615620A (zh) * 2014-06-24 2015-05-13 腾讯科技(深圳)有限公司 地图搜索类型识别方法及装置、地图搜索方法及系统
CN106095780A (zh) * 2016-05-26 2016-11-09 达而观信息科技(上海)有限公司 一种基于位置特征的检索方法
CN106933947A (zh) * 2017-01-20 2017-07-07 北京三快在线科技有限公司 一种搜索方法及装置、电子设备

Family Cites Families (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100861652B1 (ko) * 2006-11-17 2008-10-02 엔에이치엔(주) 지도 서비스 시스템 및 방법
TW201013429A (en) * 2008-09-17 2010-04-01 Yin-Kai Huang Representing method for internet geographic data target designations sorted by distance
US8145623B1 (en) * 2009-05-01 2012-03-27 Google Inc. Query ranking based on query clustering and categorization
CN102147261B (zh) * 2010-12-22 2013-07-03 广州通易科技有限公司 一种交通车辆gps数据地图匹配的方法与系统
CN103106220B (zh) * 2011-11-15 2016-08-03 阿里巴巴集团控股有限公司 一种搜索方法、搜索装置及一种搜索引擎系统
CN103123628B (zh) * 2011-11-21 2015-10-21 腾讯科技(深圳)有限公司 地理位置搜索方法及系统
CN103488654B (zh) * 2012-06-14 2017-06-16 腾讯科技(深圳)有限公司 一种基于地图进行信息搜索的搜索结果处理方法和装置
IL224482B (en) * 2013-01-29 2018-08-30 Verint Systems Ltd System and method for keyword spotting using representative dictionary
GB2532030A (en) * 2014-11-05 2016-05-11 Ibm Method and facility for user navigation in a target portal
US10872111B2 (en) * 2015-01-14 2020-12-22 Lenovo Enterprise Solutions (Singapore) Pte. Ltd User generated data based map search
CN104834721A (zh) * 2015-05-12 2015-08-12 百度在线网络技术(北京)有限公司 基于位置的搜索处理方法及装置
CN105426387B (zh) * 2015-10-23 2020-02-07 北京锐安科技有限公司 一种基于K-means算法的地图聚合方法
CN105404680A (zh) * 2015-11-25 2016-03-16 百度在线网络技术(北京)有限公司 一种搜索推荐方法及装置
CN105956181A (zh) * 2016-05-31 2016-09-21 北京百度网讯科技有限公司 搜索方法及装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103902694A (zh) * 2014-03-28 2014-07-02 哈尔滨工程大学 基于聚类和查询行为的检索结果排序方法
CN104615620A (zh) * 2014-06-24 2015-05-13 腾讯科技(深圳)有限公司 地图搜索类型识别方法及装置、地图搜索方法及系统
CN106095780A (zh) * 2016-05-26 2016-11-09 达而观信息科技(上海)有限公司 一种基于位置特征的检索方法
CN106933947A (zh) * 2017-01-20 2017-07-07 北京三快在线科技有限公司 一种搜索方法及装置、电子设备

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110874385A (zh) * 2018-08-10 2020-03-10 阿里巴巴集团控股有限公司 数据处理方法、装置和系统
CN110874385B (zh) * 2018-08-10 2023-11-14 阿里巴巴集团控股有限公司 数据处理方法、装置和系统

Also Published As

Publication number Publication date
CA3078148C (en) 2024-11-12
TWI669619B (zh) 2019-08-21
TW201828122A (zh) 2018-08-01
CN106933947B (zh) 2018-12-04
CA3078148A1 (en) 2018-07-26
CN106933947A (zh) 2017-07-07

Similar Documents

Publication Publication Date Title
WO2018133648A1 (zh) 搜索方法、装置及非临时性计算机可读存储介质
US10043520B2 (en) Multilevel speech recognition for candidate application group using first and second speech commands
US8868223B1 (en) Positioning using audio recognition
JP6759844B2 (ja) 画像を施設に対して関連付けるシステム、方法、プログラム及び装置
CN107577687B (zh) 图像检索方法及装置
WO2017166647A1 (zh) 一种导航信息推送方法和装置、设备
CN105874452B (zh) 从社交摘要中标记兴趣点
US10810374B2 (en) Matching a query to a set of sentences using a multidimensional relevancy determination
CN105404627B (zh) 一种用于确定搜索结果的方法与设备
WO2016206196A1 (zh) 用于获取用户属性信息的方法、装置及服务器
WO2017166648A1 (zh) 一种导航路线的生成方法和装置、设备
TW201604698A (zh) 一種軌跡資訊的推送方法及裝置
CN115129864B (zh) 文本分类方法、装置、计算机设备和存储介质
CN112149005B (zh) 用于确定搜索结果的方法、装置、设备和可读存储介质
WO2018176913A1 (zh) 搜索方法、装置及非临时性计算机可读存储介质
US20140280053A1 (en) Contextual socially aware local search
Dalvi et al. Deduplicating a places database
CN112528145A (zh) 信息推荐方法、装置、设备及可读存储介质
CN105653546A (zh) 一种目标主题的检索方法和系统
CN109635004A (zh) 一种数据库的对象描述提供方法、装置及设备
KR101747532B1 (ko) 여행성 질의에 대응하는 검색 결과로 코스를 추천하는 방법 및 시스템
CN109241360B (zh) 组合字符串的匹配方法及装置和电子设备
JP2020016960A (ja) 推定装置、推定方法及び推定プログラム
CN106339492A (zh) 一种基于地理位置信息的音乐推荐方法
CN115455294A (zh) 标题核心内容确定方法、搜索请求处理方法及相关装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17892143

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17892143

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 3078148

Country of ref document: CA