WO2019137365A1

WO2019137365A1 - Method and device for creating index and performing search in cloud search platform

Info

Publication number: WO2019137365A1
Application number: PCT/CN2019/070820
Authority: WO
Inventors: 葛俊
Original assignee: 阿里巴巴集团控股有限公司
Priority date: 2018-01-12
Filing date: 2019-01-08
Publication date: 2019-07-18
Also published as: TWI676112B; CN108280156A; TW201931171A

Abstract

Disclosed are a method and device for creating an index and performing a search in a cloud search platform, where the cloud search platform comprises a search instance for use by multiple subscribers, and the search instance assigns subscriber identifiers respectively to the multiple subscribers. The method for creating an index comprises the following steps: acquiring a subscriber file, the content of the subscriber file comprising a subscriber identifier field row, and the subscriber identifier field row showing a subscriber identifier of a subscriber to whom the subscriber file belongs; acquiring a subscriber dictionary of the subscriber via the subscriber identifier; performing word segmentation with respect to the subscriber file on the basis of the subscriber dictionary so as to acquire a word-segmented file corresponding to the subscriber file; and, in the search instance, creating an index with respect to the subscriber file on the basis of the field definition table and of the word-segmented file, comprising creating an index relation between the subscriber identifier field and the subscriber file on the basis of the description in the field definition table with respect to the subscriber identifier field.

Description

Method and device for constructing index and searching in cloud search platform

Technical field

The present invention relates to the field of cloud search platforms, and more particularly to a method and apparatus for constructing an index and performing a search in a cloud search platform.

Background technique

In the cloud search platform, the search function is sold externally. Each user who purchases the service is called a tenant. The number of tenants is on the 10,000 level. Different tenants have different dictionary customization requirements, and different tenants may require indexing based on different dictionaries. The traditional multi-tenant cloud search service is implemented by managing each tenant's search as a single instance, for example, 100 tenants and 100 instances. The schema (field definition table) of each instance has the same structure, but has its own corresponding word segment dictionary and configuration. Therefore, there is a need for a more efficient solution for building indexes and searching in a cloud search platform.

Summary of the invention

The embodiments of the present specification aim to provide a more efficient solution for building an index and performing a search in a cloud search platform to solve the deficiencies in the prior art.

To achieve the above object, an aspect of the present specification provides a method for constructing an index in a cloud search platform, the cloud search platform including a search instance for a plurality of tenants, the search instance including being applicable to the plurality of tenants A uniform field definition table of each tenant, the search instance is assigned a tenant ID for each tenant, the tenant ID is used to uniquely identify the corresponding tenant, and the field definition table includes a description of the tenant identification field. The tenant identification field is associated with the tenant identifier, and the method is performed by the cloud search platform and includes the following steps: acquiring a tenant document, where the content of the tenant document includes a tenant identification field row, and the tenant identification field is displayed And obtaining a tenant identifier of the tenant to which the tenant document belongs; obtaining a tenant dictionary of the tenant by using the tenant identifier; and performing segmentation according to the tenant dictionary to obtain a word segmentation document corresponding to the tenant document; And in the search instance, the rent is based on the field definition table and the word segmentation document Indexing documents, including, according to the description of the tenant identification field of the field definition table, indexing the tenant relationship with the tenant identification field document.

In one embodiment, in the above method for constructing an index in a cloud search platform, the cloud search platform further includes a dictionary unit, the dictionary unit is separated from the search instance, and the dictionary unit includes the plurality of a tenant's respective tenant dictionary, the method further comprising, after obtaining the tenant document, transmitting the tenant document to the dictionary unit.

In an embodiment, in the method for constructing an index in the cloud search platform, the tenant document is segmented according to the tenant dictionary, so that the segmented document corresponding to the tenant document is obtained: in the dictionary In the unit, the tenant document is segmented according to the tenant dictionary to generate a word segmentation document corresponding to the tenant document; and the word segmentation document is received from the dictionary unit.

In an embodiment, the method for constructing an index in the cloud search platform further includes: after segmenting the tenant document according to the tenant dictionary, thereby acquiring a word segmentation document corresponding to the tenant document, The tenant document and its corresponding word segmentation document are sent to the search instance.

In an embodiment, in the foregoing method for building an index in a cloud search platform, the cloud search platform further includes a unified service interface, the service interface is connected to the tenant platform of the plurality of tenants, and wherein the The tenant document includes: receiving, by the service interface, the tenant original document from the tenant platform, acquiring the tenant identifier according to the tenant platform, and adding the tenant identification field row to the content of the tenant original document, thereby obtaining The tenant document.

In an embodiment, in the foregoing method for building an index in a cloud search platform, the cloud search platform further includes a unified service interface, the service interface is connected to the tenant platform of the plurality of tenants, and offline The method, and the method further includes: receiving, by the service interface, a tenant original document from the tenant platform before acquiring the tenant document, and acquiring the tenant identifier according to the tenant platform, in the tenant original document The tenant identification field row is added to the content, thereby generating the tenant document, and storing the tenant document in the cloud search platform.

In an embodiment, in the method for constructing an index in the cloud search platform, the cloud search platform further includes a unified service interface, the service interface is connected to the tenant platform of the plurality of tenants, and the method The method further includes: before acquiring the tenant document, receiving a tenant dictionary from the tenant platform through the service interface, acquiring the tenant identifier according to the tenant platform, and storing the tenant dictionary in association with the tenant identifier In the dictionary unit.

Another aspect of the present specification provides an apparatus for constructing an index in a cloud search platform, the cloud search platform including a search instance for a plurality of tenants, the search instance including being applicable to each of the plurality of tenants a unified field definition table, where the search instance is assigned a tenant identifier for each tenant, the tenant identifier is used to uniquely identify the corresponding tenant, and the field definition table includes a description of the tenant identification field, and the tenant identification field In association with the tenant identifier, the device is implemented by the cloud search platform and includes the following unit: a first obtaining unit configured to acquire a tenant document, where the content of the tenant document includes a tenant identification field row, the tenant The identifier field row shows the tenant identifier of the tenant to which the tenant document belongs; the second obtaining unit is configured to acquire the tenant dictionary of the tenant by using the tenant identifier; and the word segmentation unit is configured to: according to the tenant dictionary The tenant document performs word segmentation to obtain a word segmentation document corresponding to the tenant document; and establishing a unit, configuring In the search example, the tenant document is indexed according to the field definition table and the word segmentation document, including: establishing, according to the description of the tenant identification field in the field definition table, The index relationship between the tenant identification field and the tenant document.

Another aspect of the present specification provides a method of performing a search in a cloud search platform, the cloud search platform including a search instance for a plurality of tenants, the search instance including a unified field definition applicable to the plurality of tenants The search instance is configured with a tenant identifier for each of the plurality of tenants, the tenant identifier is used to uniquely identify the corresponding tenant, and the field definition table includes a description of the tenant identification field, the tenant identification field and the tenant Identifying the association, the cloud search platform further includes a unified service interface, the service interface being connected to the tenant platform of the plurality of tenants, the method being executed by the cloud search platform and comprising the steps of: receiving a search from the tenant platform Obtaining a tenant identifier of the tenant from the tenant platform; acquiring a tenant dictionary of the tenant by using the tenant identifier; and performing word segmentation according to the tenant dictionary to obtain a word segment corresponding to the search sentence a statement; searching the tenant identification field and the segmented statement in the search instance to The tenant tenant document retrieval word have the sentence; according to the tenant identification positioning the tenant platform; and returns a search result to the tenant platform according to the field definition table.

In one embodiment, in the above method for searching in a cloud search platform, the cloud search platform further includes a dictionary unit, the dictionary unit is separated from the search instance, and the dictionary unit includes the plurality of The tenant's respective tenant dictionary, the method further comprising, after obtaining the tenant's tenant identification according to the tenant platform, transmitting the search statement and the tenant identification to the dictionary unit.

In an embodiment, in the foregoing method for searching in the cloud search platform, the word segmentation is performed according to the tenant dictionary, so that the segmented sentence corresponding to the search sentence is obtained, including: in the dictionary In the unit, the search sentence is segmented by the tenant dictionary to generate a word segmentation statement; and the word segmentation statement is received from the dictionary unit.

In one embodiment, the method for searching in the cloud search platform further includes: after the word segmentation is performed according to the tenant dictionary, to obtain the word segmentation statement corresponding to the search sentence, The word segmentation statement and the tenant identification are sent to the search instance.

Another aspect of the present specification provides an apparatus for performing a search in a cloud search platform, the cloud search platform including a search instance for a plurality of tenants, the search instance including being applicable to each of the plurality of tenants a unified field definition table, where the search instance is assigned a tenant identifier for each tenant, the tenant identifier is used to uniquely identify the corresponding tenant, and the field definition table includes a description of the tenant identification field, and the tenant identification field In association with the tenant identifier, the cloud search platform further includes a unified service interface, the service interface is connected to the tenant platform of the plurality of tenants, and the device is implemented by the cloud search platform and includes the following unit: a receiving unit, configured to receive a search statement from the tenant platform; the first obtaining unit is configured to obtain a tenant identifier of the tenant from the tenant platform; and the second obtaining unit is configured to obtain the location by using the tenant identifier a tenant dictionary of a tenant; a word segmentation unit configured to segment the search sentence according to the tenant dictionary, thereby Taking the word segmentation statement corresponding to the search statement; the retrieval unit is configured to retrieve the tenant identification field and the segmented sentence in the search instance to be in the tenant's tenant document The segmentation statement is searched; the positioning unit is configured to locate the tenant platform according to the tenant identifier; and the returning unit is configured to return a search result to the tenant platform according to the field definition table.

In the above method and apparatus for constructing an index and performing a search in a cloud search platform according to an embodiment of the present specification, a plurality of tenants are processed in a unified logic in a single search instance, and the tenant dictionary is independently external to the search instance. The service simplifies the complexity of the entire architecture, reduces development costs, and reduces the storage space required for the entire architecture. In addition, the solution can realize the requirement of multi-tenant custom dictionary, and can increase the linear growth of the number of tenants by increasing the computing resources such as servers, that is, the system is "linearly scalable".

DRAWINGS

The embodiments of the present specification can be more clearly understood by describing the embodiments of the specification with reference to the accompanying drawings:

FIG. 1 is a schematic view showing an application scenario of an embodiment of the present specification;

2 illustrates a method of building an index in a cloud search platform in accordance with one embodiment of the present specification;

FIG. 3 illustrates an apparatus 300 for building an index in a cloud search platform in accordance with an embodiment of the present specification;

4 illustrates a method of performing a search in a cloud search platform in accordance with an embodiment of the present specification;

FIG. 5 illustrates an apparatus 500 for performing a search in a cloud search platform in accordance with an embodiment of the present specification.

Detailed ways

Embodiments of the present specification will be described below with reference to the drawings.

FIG. 1 schematically shows an application scenario of an embodiment of the present specification. The application scenarios of the embodiments of the present specification include a cloud search platform 101 and tenant platforms 102, 103, 104, and the like of a plurality of tenants. The cloud search platform 101 includes a unified service interface for connecting to various tenant platforms, such as a restful service interface. For example, the cloud search platform 101 can receive a tenant's document, a tenant's custom dictionary, etc. from the tenant platform 103 through a service interface, thereby building an index from the tenant's document and a custom dictionary. When the user of the tenant platform 103 searches on the tenant platform 103 using the search engine, the tenant platform 103 sends its user's search statement to the cloud search platform 101 through the service interface. The cloud search platform 101 searches by distinguishing tenants by a field line such as "tenant ID = xxx" (where xxx is the tenant's tenant identification), and returns the search result to the tenant platform 103.

The cloud search platform includes a search instance in which a plurality of tenants are included, the search instances including a unified field definition schema applicable to each of the plurality of tenants. That is, multiple tenants share a single search instance, using the same field definition table. The search example here is a stand-alone application in the cloud search platform for implementing the search function, and the search instance and the search instance are logically isolated from each other.

The field definition table (schema) can for example be in the form of a schema.xml configuration file containing all the fields that the document may contain and all the information about how these fields will be processed when creating the document index and query.

For example, the document of the original data (ie, raw doc) is generated in the following format:

Id=1

User_id=001

Title=xxxxxxx yyyyy

Content=...origin_title=xxxxxyyyyy

The id, user_id, title, and content are the fields included in the document, and the content after the “=” is the value of the corresponding field.

Table 1 schematically shows a field definition table for a search instance.

Table 1

Table 1 lists the fields included in the document processed in the search instance: title (title after word segmentation), content (body), cat_id (category id), user_id (tenant id), origin_title (original title). Table 1 also records information about how these fields will be processed when creating document indexes and queries. For example, in the title line, the "inverted, positive row" in the column of the engine field indicates that the inverted index and the positive index will be created for the title, and that the index is displayed in the column if the word segmentation is required. Spaces are used to segment the title. For another example, in the content line, the "summary field" in the column of the engine field indicates that the body is displayed in the form of a digest when the query result is displayed. Of course, the field definition table in Table 1 is only exemplary, but is used to illustrate the field definition table, and the field definition table in the actual application can contain more fields and can contain different field definitions.

The search instance reads the above field definition table when building an index for each tenant's document and builds an index based on the description in the table. In addition, when returning search results based on a tenant search request, the search instance also displays the fields according to the description in the field definition table.

The search instance is that each of the plurality of tenants is assigned a tenant identifier, and the tenant identifier is used to uniquely identify the corresponding tenant. Specifically, the search instance distinguishes different tenants by one field (for example, user_id) in the field definition table. For example, "user_id=001" indicates that it is the first tenant, and "user_id=002" indicates the second tenant. The tenant ID can also be used to distinguish tenant's documents, tenant's dictionary, tenant platform, and so on. For example, by associating a tenant's dictionary with a tenant ID, when using a tenant dictionary, the tenant dictionary can be obtained through the tenant ID. When searching in the search instance, by distinguishing the requests of different tenants by including "user_id=xxx" in the search request, it is possible to achieve mutual interference between the tenants. For example, the specific search command (query) can be: query=title: weather AND user_id: xxx, the meaning of this search sentence is to search for the index containing "weather" in the title, and the index needs to satisfy the condition that the user_id is xxx. This is equivalent to filtering by tenant dimension.

As can be seen from the above description, the field definition tables used by multiple tenants in a search instance are consistent, but the dictionaries of different tenants are different. For example, some sports articles require the use of sports proprietary lexicons, some The medical class requires the use of a medical dictionary. The search instance classifies the tenant's original document according to the tenant's respective dictionary when building the tenant's index for building the index. In one embodiment of the present specification, the tenant-related word segmentation logic is implemented outside the search instance, so that the search instance only performs general storage, indexing, retrieval, and the like, without complicated tenant dictionary customization.

Specifically, as shown in FIG. 1, the cloud search platform 101 further includes a dictionary unit 12, a service proxy unit 13, and a storage unit 14. The dictionary unit 12 is separated from the search instance 11, and the dictionary unit includes respective tenant dictionaries of the plurality of tenants, which can provide word segmentation services to individual tenants in addition to the search instances. Specifically, when the service of the dictionary unit 12 is invoked for a document from the tenant platform 102, the dictionary unit 12 uses the tenant's tenant dictionary to segment the tenant's document.

The service agent unit 13 is configured to proxy the service of the cloud search platform 101 and provide an http service upward, which is respectively connected with the tenant platform (102, 103, 104, etc.), the search instance 11, the dictionary unit 12, the storage unit 14, etc., to Transfer data between them and perform appropriate data preprocessing. For example, in the case of offline indexing, the service agent unit 13 receives the original document of the tenant from the tenant platform 103, adds the tenant identification to the original document, obtains the tenant document, and stores the tenant document in the storage unit 14. Upon construction of the index, the service agent unit 13 acquires the tenant document from the storage unit 14, transmits the document to the dictionary unit 12 for word segmentation, and receives the word segmentation document from the dictionary unit 12. The business agent unit 13 then sends the tenant document and the word segmented document to the search instance 11, which searches the tenant document according to the field definition table and the tenant's word segmentation document, wherein the index includes the field index of the tenant ID.

The configuration of the cloud search platform 101 shown in FIG. 1 is only one embodiment of the present specification, and does not limit the embodiments of the present specification. In another embodiment, in the cloud search platform, the dictionary word segmentation function is set inside the search instance. Thus, the tenant document is only sent to the search instance, and the tenant document is segmented by the search instance according to the tenant dictionary. In another embodiment, in the cloud search platform, the search instance calls the service of the external dictionary unit directly (ie, not through the service agent unit), that is, directly obtains the word segmentation document and the tenant document from the dictionary unit for construction. index.

A method and apparatus for constructing an index in a cloud search platform in accordance with one embodiment of the present specification is described below. 2 illustrates a method of building an index in a cloud search platform in accordance with one embodiment of the present specification.

As shown in FIG. 2, in step S21, a tenant document is acquired. The content of the tenant document includes a tenant identification field row, and the tenant identification field row indicates a tenant identifier of a tenant to which the tenant document belongs. As described above, the cloud search platform includes a unified service interface, and the service interface is connected to the tenant platforms of the plurality of tenants. The tenant original document is received from the tenant platform through the service interface.

For example, in the case of constructing an index in real time, the tenant original document is received from the tenant platform through the service interface, and the tenant identifier “xxx” is obtained according to the tenant platform, and a field row is added to the content of the tenant original document. "Tenant ID = xxx" to obtain the tenant document. For example, when the cloud search platform receives the tenant original document from the tenant platform of the tenant 001, the parameter such as type=user_001 is also received from the tenant platform, so that the tenant's tenant identifier “xxx”, for example, “001” can be obtained. In the case of offline indexing, the tenant original document is received from the tenant platform through the service interface, and the tenant identifier “xxx” is obtained according to the tenant platform, and a field row is added to the content of the tenant original document. The tenant ID = xxx" line, thereby generating the tenant document and storing the tenant document in the cloud search platform. Thus, when the index is built offline, the tenant document is obtained from the storage unit of the cloud search platform.

In one embodiment, as shown in FIG. 1, a service agent unit is included in the cloud search platform for proxying business logic in the platform. The unified service interface is set on the service agent unit, and the service interface is connected to tenant platforms of multiple tenants. And the service agent unit is further connected to the storage unit of the cloud search platform. For example, in the case of constructing an index in real time, the service agent unit receives a tenant original document from the tenant platform through the service interface, and acquires a tenant identification “xxx” according to the tenant platform, in the tenant original document The field line "tenant ID=xxx" is added to the content to obtain the tenant document. In the case of offline indexing, the service agent unit receives the tenant original document from the tenant platform through the service interface, and obtains the tenant identifier “xxx” according to the tenant platform, and the content of the original document in the tenant. The field row "tenant ID=xxx" row is added to generate the tenant document, and the tenant document is stored in the storage unit 14 of the cloud search platform. Thus, when the index is built offline, the service agent unit acquires the tenant document from the storage unit 14.

In step S22, the tenant dictionary of the tenant is obtained by the tenant identifier. The cloud search platform further receives a tenant dictionary from the tenant platform through the service interface, acquires a tenant identifier according to the tenant platform, and stores the tenant dictionary in the dictionary unit in association with the tenant identifier. Thus, when the tenant document is indexed, the tenant dictionary associated with the tenant identification can be proposed from the platform by the tenant identification, and the tenant dictionary can be obtained.

In step S23, the tenant document is segmented according to the tenant dictionary, thereby acquiring a word segmentation document corresponding to the tenant document. For example, the title of the original document is “Beijing Group Buying Website Long”, and the tenant dictionary of tenant 001 includes the terms “group purchase” and “website”, and the result of the word segmentation according to the tenant dictionary of tenant 001 is “Beijing group purchase website”. long". For another example, if the tenant dictionary of the tenant 002 includes the terms "Beijing Group Buying Network" and "Webmaster", the result of the word segmentation according to the tenant dictionary of the tenant 002 is "Beijing Group Buying Website Long". In the above word segmentation results, spaces are used as the separation between the word segments, which is merely exemplary, and may be separated by other characters or expressed in other forms, for example, by structuring each word segmentation.

In one embodiment, the tenant document may be further segmented using the default dictionary in the dictionary unit. In this case, the terms in the tenant dictionary are used preferentially.

In one embodiment, as shown in FIG. 1, a dictionary unit is further included in the cloud search platform, and the dictionary unit is an application on the platform separate from the search instance, and is used to provide a word segmentation service.

The service agent unit receives the tenant dictionary from the tenant platform through the service interface, acquires the tenant ID according to the tenant platform, and stores the tenant dictionary in the dictionary unit in association with the tenant identification. For example, you can set up a startup configuration file in the tenant dictionary. One option of the configuration file is key: dict_path, for example, user_id_001:/home/admin/local_dict_1.txt, so that the tenant dictionary is associated with the tenant ID through the configuration file. stand up.

Thus, when an index is built for a tenant, the word segmentation service of the dictionary unit can be called by the business agent unit. For example, the business agent unit sends the tenant document to the dictionary unit after acquiring the tenant document. The dictionary unit obtains the tenant ID through the field line "tenant ID=xxx" in the tenant document, and proposes a tenant dictionary stored in association with the tenant ID from the tenant ID, and acquires the tenant dictionary. Then, in the dictionary unit, the tenant document is segmented according to the tenant dictionary to generate a word segmentation document. Thereafter, the dictionary unit transmits the tenant document and the segmented document to the service agent unit.

In step S24, in the search instance, the tenant document is indexed according to the field definition table and the word segmentation document, including: according to the description of the tenant identification field in the field definition table Establishing an index relationship between the tenant identification field and the tenant document. For example, the field definition table in the search instance is a table as shown in Table 1, which shows the inverted index and the positive row for the title, title id (cat_id), tenant ID (user_id) of the word segmentation. index. Taking the tenant ID as an example, according to the description of the tenant ID field in the field definition table, the indexer in the search instance generates an index relationship table of the tenant ID and the tenant document, including the inverted list and the positive list.

As shown in FIG. 1, in one embodiment, after the tenant document is segmented according to the tenant dictionary in the dictionary unit 12, and the tenant document and the word segmented document are transmitted to the service agent unit 13, the service agent unit The tenant document and the word segmented document are sent to the search instance 11. The content of the word segmentation can be separated by a pre-agreed separator (for example, a space). In the case where the document is not segmented by the separator in the dictionary unit, the word segmentation document can be further processed in the business agent unit. Modify the document to be segmented with the convention separator. In this case, the search instance performs word segmentation according to the completed word segmentation field after receiving the word segmentation document, for example, word segmentation according to spaces, without additional word segmentation processing.

Thus, in the search instance, all tenant documents can be processed in a unified logic without the need to distinguish tenants. That is, the index of the search instance includes the documents of all the tenants, and when searching, the requests of different tenants are distinguished by adding, for example, user_id=“xxx”, so that the tenants can be isolated.

FIG. 3 illustrates an apparatus 300 for building an index in a cloud search platform in accordance with an embodiment of the present specification. The cloud search platform includes a search instance for a plurality of tenants, the search instance including a unified field definition table applicable to the plurality of tenants, the search instance each having a tenant identifier assigned to each of the tenants, The tenant identifier is used to uniquely identify the corresponding tenant, and the field definition table includes a description of the tenant identification field, and the tenant identification field is associated with the tenant identifier.

As shown in FIG. 3, the device 300 for constructing an index in the cloud search platform is implemented by the cloud search platform and includes the following unit: a first obtaining unit 31 configured to acquire a tenant document, where the content of the tenant document includes a tenant An identifier field row, the tenant identification field row showing a tenant identifier of the tenant to which the tenant document belongs; the second obtaining unit 32 is configured to obtain a tenant dictionary by using the tenant identifier, and the word segmentation unit 33 is configured to a tenant dictionary segmenting the tenant document to obtain a word segmentation document corresponding to the tenant document; and an establishing unit 34 configured to, according to the field definition table, the word segmentation document The tenant document is indexed, and the index relationship between the tenant identification field and the tenant document is established according to the description of the tenant identification field in the field definition table.

In one embodiment, the cloud search platform further includes a dictionary unit, the dictionary unit is separated from the search instance, and the dictionary unit includes respective tenant dictionaries of the plurality of tenants, the cloud search The apparatus 300 for constructing an index in the platform further includes a first sending unit configured to send the tenant document to the dictionary unit after acquiring the tenant document.

In one embodiment, the segmentation of the tenant document according to the tenant dictionary, so as to obtain the word segmentation document corresponding to the tenant document, includes: in the dictionary unit, the tenant document according to the tenant dictionary Word segmentation is performed to generate a word segmentation document corresponding to the tenant document; and the word segmentation document is received from the dictionary unit.

In an embodiment, the apparatus 300 for constructing an index in the cloud search platform further includes a second sending unit configured to perform segmentation on the tenant document according to the tenant dictionary, thereby acquiring the corresponding corresponding to the tenant document After the word segmentation document, the tenant document and its corresponding word segmentation document are sent to the search instance.

In an embodiment, in the apparatus 300 for building an index in a cloud search platform, the cloud search platform further includes a unified service interface, the service interface is connected to the tenant platform of the plurality of tenants, and The obtaining the tenant document includes: receiving, by the service interface, the tenant original document from the tenant platform, obtaining the tenant identifier according to the tenant platform, and adding the tenant identification field row to the content of the tenant original document. Thereby obtaining the tenant document.

In one embodiment, the apparatus 300 for constructing an index in the cloud search platform is implemented offline, and the apparatus 300 further includes: a first storage unit configured to pass through the service interface before acquiring the tenant document The tenant platform receives the tenant original document, obtains the tenant identifier according to the tenant platform, adds the tenant identification field row to the content of the tenant original document, thereby generating the tenant document, and the tenant document Stored in the cloud search platform.

In an embodiment, the apparatus 300 for constructing an index in the cloud search platform further includes a second storage unit configured to receive a tenant dictionary from the tenant platform through the service interface before acquiring the tenant document, according to the The tenant platform obtains the tenant identification and stores the tenant dictionary in the dictionary unit in association with the tenant identification.

FIG. 4 illustrates a method of performing a search in a cloud search platform in accordance with an embodiment of the present specification. The cloud search platform includes a search instance for a plurality of tenants, the search instance including a unified field definition table applicable to the plurality of tenants, the search instance each having a tenant identifier assigned to each of the ten tenants, The tenant identifier is used to uniquely identify the corresponding tenant, the field definition table includes a description of the tenant identification field, the tenant identification field is associated with the tenant identifier, and the cloud search platform includes a unified service interface, The service interface is connected to the tenant platform of the plurality of tenants.

As shown in FIG. 4, in step S41, a search sentence is received from the tenant platform. The cloud search platform receives the tenant's search request from the tenant platform through the unified service interface described above. For example, the search request is the search phrase "Beijing Group Buy".

In one embodiment, as shown in FIG. 1, the cloud search platform 101 includes a service proxy unit 13 that includes the service interface to connect with the tenant platform. The service agent unit 13 receives the tenant's search request from the tenant platform through the unified service interface described above.

In step S42, the tenant's tenant identification is obtained from the tenant platform. For example, each tenant platform may include a tenant identification parameter corresponding to the tenant platform in the request string it sends. For example, the tenant platform of tenant 001 sends a parameter such as type=user_001 to the cloud search platform. Thus, from the parameters included in the request string by the tenant platform, the tenant's tenant identification "xxx", for example, "001" can be obtained. In one embodiment, as shown in Figure 1, the tenant's tenant identification "xxx" is obtained from the tenant platform via the business agent layer 13.

In step S43, the tenant dictionary of the tenant is obtained by the tenant identifier. As described above, the tenant dictionary is stored in the cloud search platform in association with the tenant identification, so that the tenant dictionary associated with the tenant identification can be proposed from the platform through the tenant identification, and the tenant dictionary can be obtained.

In one embodiment, as shown in FIG. 1, cloud search platform 101 also includes a dictionary unit 12. The business agent unit 13 transmits the search sentence and the tenant identification to the dictionary unit 12 after acquiring the tenant's tenant identification "xxx" from the tenant platform. Thus, the dictionary unit 12 proposes a tenant dictionary associated with the tenant identification from itself through the tenant identification, and acquires the tenant dictionary.

In step S44, the search sentence is segmented according to the tenant dictionary, thereby acquiring a word segmentation sentence corresponding to the search sentence. For example, when the cloud search platform receives the search sentence "Beijing Group Purchase" from the tenant platform of the tenant 001, the tenant dictionary of the tenant 001 includes the terms "group purchase" and "website", and the sentence is segmented according to the tenant dictionary of the tenant 001. The result is "Beijing group purchase." For another example, when the cloud search platform receives the search sentence “Beijing Group Purchase” from the tenant platform of the tenant 002, the tenant dictionary of the tenant 002 includes the term “Beijing Group Purchase Network”, and the sentence is segmented according to the tenant dictionary of the tenant 002. The result is still "Beijing group purchase."

In one embodiment, as shown in FIG. 1, in the dictionary unit 12, the search sentence is segmented by the tenant dictionary to generate a word segmentation statement. Thereafter, the dictionary unit 12 sends the word segmentation statement and the tenant identification back to the service agent unit 13. Thereafter, the service agent unit 13 transmits the word segmentation statement and the tenant identification to the search instance 11.

In step S45, the tenant identification field and the segmented sentence are retrieved in the search instance to retrieve the word segmentation statement in the tenant's tenant document.

For example, the document of tenant 001 (ie, tenant ID=001) includes the document titled “Beijing Group Buying Website Long”, and the tenant dictionary of tenant 001 includes the terms “group purchase” and “website”, then the tenant according to tenant 001 The result of the dictionary segmentation of the title is “Beijing Group Buying Website Long”, that is, the index entry corresponding to the document of tenant 001 includes “Beijing”, “Group Purchase”, “Website” and “Long”. When the "Beijing Group Purchase" search sentence is received from the tenant platform of the tenant 001, the word segmentation sentence that is segmented by the tenant dictionary of the tenant 001 is "Beijing Group Purchase". Both "Beijing" and "Group Buy" are associated with the document titled "Beijing Group Buying Site Long" in the index, and the document is also associated with "tenant ID=001" in the index (ie, the document is tenant 001) Document), so in this case, the search sentence "Beijing Group Buy" is searched by the search instance, and the document titled "Beijing Group Buying Website Long" will be returned.

For another example, the document of tenant 002 (ie, tenant ID=002) includes a document titled “Beijing Group Buying Website Long”, and the tenant dictionary of tenant 002 includes the terms “Beijing Group Buying Network” and “Webmaster”, according to The result of the segmentation of the title by the tenant dictionary of tenant 002 is “Beijing Group Buying Website Long”, that is, the index entry corresponding to the document of tenant 002 includes “Beijing Group Buying Network” and “Webmaster”. When the "Beijing Group Purchase" search sentence is received from the tenant platform of the tenant 002, the word segmentation sentence that is segmented by the tenant dictionary of the tenant 002 is still "Beijing Group Purchase". "Beijing Group Buy" is not associated with the document titled "Beijing Group Buying Website Long" in the index, so that in this case, the search sentence "Beijing Group Buy" is searched by the search instance, and the title will not be returned. The document of "Beijing Group Buying Website Director".

In one embodiment, as shown in FIG. 1, after searching in the search example 11 according to the field row "tenant ID=xxx" and the word segmentation statement, the search instance 11 will display the search result according to the field definition table and The tenant identification is sent to the service agent unit 13.

In step S46, the tenant platform is located according to the tenant identification. In the embodiment of the present specification, the cloud search platform is connected to the tenant platforms of multiple tenants through a unified service interface, so that when the result is returned, the platform can locate the tenant platform before sending the search request through the tenant identification.

In one embodiment, as shown in FIG. 1, the service agent unit 13 locates the tenant platform that previously sent the search request by the tenant identification received from the search instance 11.

Finally, in step S47, the search result is returned to the tenant platform according to the field definition table. For example, referring to the field definition table shown in Table 1 above, the "engine field" column in the table defines the content line as "summary field", that is, when the search result is returned, the content is displayed in the summary field. According to the table, it is also found that cat_id (category id) and origin_title (original title) are also displayed in the summary.

In one embodiment, as shown in FIG. 1, the service agent unit 13 returns the search result displayed according to the field definition table received from the search instance 11 to the tenant platform after the tenant platform is located.

FIG. 5 illustrates an apparatus 500 for performing a search in a cloud search platform in accordance with an embodiment of the present specification. The cloud search platform includes a search instance for a plurality of tenants, the search instance including a unified field definition table applicable to each of the plurality of tenants, the search instance each having a tenant assigned to a tenant And the tenant identifier is used to uniquely identify the corresponding tenant, the field definition table includes a description of the tenant identification field, the tenant identification field is associated with the tenant identifier, and the cloud search platform further includes a unified service interface. The service interface is connected to the tenant platform of the plurality of tenants.

As shown in FIG. 5, the device 500 for searching in the cloud search platform is implemented by the cloud search platform and includes the following unit: a first receiving unit 51 configured to receive a search sentence from the tenant platform; The obtaining unit 52 is configured to acquire the tenant identifier of the tenant from the tenant platform. The second obtaining unit 53 is configured to acquire the tenant dictionary of the tenant by using the tenant identifier, and the word segmentation unit 54 is configured to The tenant dictionary performs segmentation on the search sentence to obtain a word segmentation statement corresponding to the search sentence; the retrieval unit 55 is configured to retrieve the tenant identification field and the segmented sentence in the search instance Retrieving the word segmentation statement in the tenant's tenant document; the positioning unit 56 is configured to locate the tenant platform according to the tenant identifier; and the returning unit 57 is configured to be defined according to the field The table returns the search results to the tenant platform.

In one embodiment, the cloud search platform further includes a dictionary unit, the dictionary unit is separated from the search instance, and the dictionary unit includes respective tenant dictionaries of the plurality of tenants, the cloud search The device 500 for searching in the platform further includes a first sending unit, configured to send the search statement and the tenant identifier to the dictionary unit after acquiring the tenant identifier of the tenant according to the tenant platform.

In an embodiment, in the device 500 for searching in the cloud search platform, the search term is segmented according to the tenant dictionary, so as to obtain the word segmentation statement corresponding to the search sentence, including: In the dictionary unit, the search sentence is segmented by the tenant dictionary to generate a word segmentation sentence; and the word segmentation sentence is received from the dictionary unit.

In an embodiment, the device 500 for searching in the cloud search platform further includes: a second sending unit, configured to: after receiving the tenant identifier from the dictionary unit, the word segmentation statement and the The tenant identification is sent to the search instance.

The embodiment of the present specification further includes a computer readable storage medium having stored thereon an instruction code, when executed in a computer, causing a computer to execute indexing and performing in a cloud search platform according to an embodiment of the present specification. The method of searching.

Those of ordinary skill in the art should further appreciate that the elements and algorithm steps of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both, in order to clearly illustrate the hardware. Interchangeability with software, the components and steps of the various examples have been generally described in terms of functionality in the above description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the solution. Different methods may be used to implement the described functionality for each particular application, but such implementation should not be considered to be beyond the scope of the application.

The steps of a method or algorithm described in connection with the embodiments disclosed herein may be implemented in hardware, in a software module in a processor orbit, or in a combination of the two. The software module can be placed in random access memory (RAM), memory, read only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or technical field. Any other form of storage medium known.

The specific embodiments of the present invention have been described in detail with reference to the preferred embodiments of the present invention. All modifications, equivalent substitutions, improvements, etc., made within the spirit and scope of the invention are intended to be included within the scope of the invention.

Claims

A method of building an index in a cloud search platform, the cloud search platform including a search instance for a plurality of tenants, the search instance including a unified field definition table applicable to the plurality of tenants, the search instance A plurality of tenants are each assigned a tenant identifier, the tenant identifier is used to uniquely identify the corresponding tenant, the field definition table includes a description of the tenant identification field, and the tenant identification field is associated with the tenant identifier, the method Executed by the cloud search platform and including the following steps:

Obtaining a tenant document, where the content of the tenant document includes a tenant identification field row, where the tenant identification field row shows a tenant identifier of a tenant to which the tenant document belongs;

Obtaining, by the tenant identifier, a tenant dictionary of the tenant;

Deriving the tenant document according to the tenant dictionary to obtain a word segmentation document corresponding to the tenant document;

In the search example, indexing the tenant document according to the field definition table and the word segmentation document, including: establishing, according to the description of the tenant identification field in the field definition table, The index relationship between the tenant identification field and the tenant document.
The method of constructing an index in a cloud search platform according to claim 1, wherein the cloud search platform further comprises a dictionary unit, the dictionary unit is separated from the search instance, and the plurality of dictionary units are included Tenant's respective tenant dictionary, the method also includes

After obtaining the tenant document, the tenant document is sent to the dictionary unit.
The method for constructing an index in a cloud search platform according to claim 2, wherein the segmentation of the tenant document according to the tenant dictionary, so as to obtain the word segmentation document corresponding to the tenant document, comprises:

In the dictionary unit, segmenting the tenant document according to the tenant dictionary to generate a word segmentation document corresponding to the tenant document;

The word segmented document is received from the dictionary unit.
The method for constructing an index in a cloud search platform according to claim 2, further comprising: after segmenting the tenant document according to the tenant dictionary, thereby acquiring a word segmentation document corresponding to the tenant document, The tenant document and its corresponding word segmentation document are sent to the search instance.
The method for constructing an index in a cloud search platform according to any one of claims 1 to 4, wherein the cloud search platform further comprises a unified service interface, the service interface and the tenant platform of the plurality of tenants. Connecting, and wherein obtaining the tenant document includes receiving, by the service interface, a tenant original document from the tenant platform, obtaining the tenant identification according to the tenant platform, and adding the tenant to content of the tenant original document Identify the field row to get the tenant document.
The method for constructing an index in a cloud search platform according to any one of claims 1 to 4, wherein the cloud search platform further comprises a unified service interface, the service interface and the tenant platform of the plurality of tenants. Connecting, and performing the method offline, and the method further includes: receiving, by the service interface, a tenant original document from the tenant platform before acquiring the tenant document, and acquiring the tenant identifier according to the tenant platform, Adding the tenant identification field row to the content of the tenant original document, thereby generating the tenant document, and storing the tenant document in the cloud search platform.
The method for constructing an index in a cloud search platform according to any one of claims 2 to 4, wherein the cloud search platform further comprises a unified service interface, the service interface and the tenant platform of the plurality of tenants. Connecting, and the method further comprises: before acquiring the tenant document, receiving a tenant dictionary from the tenant platform through the service interface, acquiring the tenant identifier according to the tenant platform, and the tenant dictionary and the tenant The identification is stored in association with the dictionary unit.
An apparatus for building an index in a cloud search platform, the cloud search platform including a search instance for a plurality of tenants, the search instance including a unified field definition table applicable to the plurality of tenants, the search instance Each tenant is assigned a tenant identifier, the tenant identifier is used to uniquely identify the corresponding tenant, the field definition table includes a description of the tenant identification field, and the tenant identification field is associated with the tenant identifier, the device Implemented by the cloud search platform and includes the following elements:

The first obtaining unit is configured to obtain a tenant document, where the content of the tenant document includes a tenant identification field row, and the tenant identification field row indicates a tenant identifier of a tenant to which the tenant document belongs;

a second acquiring unit, configured to acquire, by using the tenant identifier, a tenant dictionary of the tenant;

a word segmentation unit configured to perform segmentation on the tenant document according to the tenant dictionary, thereby acquiring a word segmentation document corresponding to the tenant document;

An establishing unit, configured to: in the search instance, indexing the tenant document according to the field definition table and the word segmentation document, including: according to the tenant identification field in the field definition table The description establishes an index relationship between the tenant identification field and the tenant document.
The apparatus for constructing an index in a cloud search platform according to claim 8, wherein the cloud search platform further comprises a dictionary unit, the dictionary unit is separated from the search instance, and the plurality of dictionary units are included Tenant's respective tenant dictionary, the device also includes

The first sending unit is configured to send the tenant document to the dictionary unit after acquiring the tenant document.
The apparatus for constructing an index in a cloud search platform according to claim 9, wherein the segmentation of the tenant document according to the tenant dictionary, so as to obtain the word segmentation document corresponding to the tenant document, comprises:

In the dictionary unit, segmenting the tenant document according to the tenant dictionary to generate a word segmentation document corresponding to the tenant document;

The word segmented document is received from the dictionary unit.
The apparatus for constructing an index in a cloud search platform according to claim 9, further comprising: a second sending unit configured to perform word segmentation according to the tenant dictionary to obtain a corresponding to the tenant document After the word segmentation document, the tenant document and its corresponding word segmentation document are sent to the search instance.
The apparatus for constructing an index in a cloud search platform according to any one of claims 8-11, wherein the cloud search platform further comprises a unified service interface, the service interface and the tenant platform of the plurality of tenants. Connecting, and wherein obtaining the tenant document includes receiving, by the service interface, a tenant original document from the tenant platform, obtaining the tenant identification according to the tenant platform, and adding the tenant to content of the tenant original document Identify the field row to get the tenant document.
The apparatus for constructing an index in a cloud search platform according to any one of claims 8-11, wherein the cloud search platform further comprises a unified service interface, the service interface and the tenant platform of the plurality of tenants. Connecting, and wherein the apparatus is implemented offline, and the apparatus further includes a first storage unit configured to receive a tenant original document from the tenant platform through the service interface before acquiring the tenant document, and according to the The tenant platform obtains the tenant identifier, adds the tenant identifier field row to the content of the tenant original document, generates the tenant document, and stores the tenant document in the cloud search platform.
The apparatus for constructing an index in a cloud search platform according to any one of claims 9-11, wherein the cloud search platform further comprises a unified service interface, the service interface and the tenant platform of the plurality of tenants. Connecting, and wherein the apparatus further includes: a second storage unit configured to receive a tenant dictionary from the tenant platform through the service interface before acquiring the tenant document, obtain the tenant identifier according to the tenant platform, and The tenant dictionary is stored in the dictionary unit in association with the tenant identification.
A method of searching in a cloud search platform, the cloud search platform including a search instance for a plurality of tenants, the search instance including a unified field definition table applicable to the plurality of tenants, the search instance Each tenant is assigned a tenant ID, the tenant identifier is used to uniquely identify the corresponding tenant, and the field definition table includes a description of the tenant identification field, where the tenant identification field is associated with the tenant identifier, the cloud The search platform also includes a unified service interface, the service interface being connected to the tenant platform of the plurality of tenants, the method being performed by the cloud search platform and comprising the following steps:

Receiving search statements from the tenant platform;

Obtaining a tenant's tenant ID from the tenant platform;

Obtaining, by the tenant identifier, a tenant dictionary of the tenant;

And segmenting the search sentence according to the tenant dictionary, thereby acquiring a word segmentation statement corresponding to the search sentence;

Retrieving the tenant identification field and the segmented sentence in the search instance to retrieve the segmented statement in the tenant's tenant document;

Locating the tenant platform based on the tenant identification;

Returning the search result to the tenant platform according to the field definition table.
The method for searching in a cloud search platform according to claim 15, wherein the cloud search platform further comprises a dictionary unit, the dictionary unit is separated from the search instance, and the plurality of dictionary units are included Tenant's respective tenant dictionary, the method also includes

After obtaining the tenant identification of the tenant according to the tenant platform, the search statement and the tenant identification are sent to the dictionary unit.
The method for performing a search in a cloud search platform according to claim 16, wherein the segmentation of the search sentence according to the tenant dictionary, so as to obtain the word segmentation statement corresponding to the search sentence, comprises:

In the dictionary unit, the search sentence is segmented by the tenant dictionary to generate a word segmentation statement;

The word segmentation statement is received from the dictionary unit.
The method for searching in a cloud search platform according to claim 16, further comprising: after segmenting the search sentence according to the tenant dictionary, thereby acquiring a word segmentation sentence corresponding to the search sentence, The word segmentation statement and the tenant identification are sent to the search instance.
An apparatus for searching in a cloud search platform, the cloud search platform including a search instance for a plurality of tenants, the search instance including a unified field definition table applicable to the plurality of tenants, the search instance Each tenant is assigned a tenant ID, the tenant identifier is used to uniquely identify the corresponding tenant, and the field definition table includes a description of the tenant identification field, where the tenant identification field is associated with the tenant identifier, the cloud The search platform also includes a unified service interface, the service interface being connected to the tenant platform of the plurality of tenants, the device being implemented by the cloud search platform and comprising the following units:

a first receiving unit, configured to receive a search statement from the tenant platform;

a first acquiring unit, configured to acquire a tenant identifier of the tenant from the tenant platform;

a second acquiring unit, configured to acquire, by using the tenant identifier, a tenant dictionary of the tenant;

a word segmentation unit configured to perform segmentation on the search sentence according to the tenant dictionary, thereby acquiring a word segmentation statement corresponding to the search sentence;

a retrieval unit configured to retrieve the tenant identification field and the segmented sentence in the search instance to retrieve the segmented statement in the tenant's tenant document;

a positioning unit configured to locate the tenant platform according to the tenant identifier;

Returning to the unit, configured to return a search result to the tenant platform according to the field definition table.
The apparatus for performing a search in a cloud search platform according to claim 19, wherein the cloud search platform further comprises a dictionary unit, the dictionary unit is separated from the search instance, and the plurality of dictionary units are included Tenant's respective tenant dictionary, the device also includes

The first sending unit is configured to send the search statement and the tenant identifier to the dictionary unit after acquiring the tenant identifier of the tenant according to the tenant platform.
The apparatus for performing a search in a cloud search platform according to claim 20, wherein the segmentation of the search sentence according to the tenant dictionary, so as to obtain the word segmentation statement corresponding to the search sentence, comprises:

In the dictionary unit, the search sentence is segmented by the tenant dictionary to generate a word segmentation statement;

The word segmentation statement is received from the dictionary unit.
The apparatus for searching in a cloud search platform according to claim 20, further comprising: a second sending unit configured to perform word segmentation according to the tenant dictionary to obtain a correspondence corresponding to the search sentence After the word segmentation statement, the word segmentation statement and the tenant identifier are sent to the search instance.