Search method and device for live broadcast room, server and storage medium

Abstract

The present disclosure relates to a live room searching method and device, a server and a storage medium. The method comprises: obtaining an input text; performing intent analysis on the input text to obtain at least one searching intent; if the searching intent includes a live searching intent, retrieving a target live room identifier from live index data corresponding to the live searching intent, wherein the live index data includes merged static data stream and dynamic data stream corresponding to the target live room identifier. The method constructs live index data according to the dynamic data stream and the static data stream corresponding to the live room identifier, and utilizes the real-time dynamic change characteristics of the live room data, thereby effectively improving the accuracy of live room searching and greatly improving the user search volume.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of Internet, and particularly relates to a search method and device for a live broadcast room, a server and a storage medium. BACKGROUND

[0002] With the development of live broadcast technology, the number of live broadcast rooms in various live broadcast applications increases rapidly. In order to enable users to find a live broadcast room of interest from a large number of live broadcast rooms, many live broadcast applications provide a search entrance for live broadcast rooms.

[0003] In the related art, live broadcast search is usually based on static data such as anchor related information and live broadcast room tags to establish an index database. The index database is used to respond to a search request for a live broadcast room proposed by a user and provide live broadcast room information that the user wants to obtain. When input text of a viewer client is obtained, the input text is used to query a live broadcast room matching the input text from the index database, and the matching live broadcast room is sent to the viewer client. However, due to the dynamic change characteristics of live broadcast room data, the above-mentioned search method for a live broadcast room has the problem of insufficient accuracy. SUMMARY

[0004] The present disclosure provides a search method and device for a live broadcast room, a server and a storage medium to at least solve the problem of insufficient accuracy of live broadcast room search in the related art. The technical solutions of the present disclosure are as follows.

[0005] According to a first aspect of an embodiment of the present disclosure, a search method for a live broadcast room is provided, comprising:

[0006] obtaining input text;

[0007] performing intent analysis on the input text to obtain at least one search intent;

[0008] if the search intent includes a live broadcast search intent, retrieving a target live broadcast room identifier from live broadcast index data corresponding to the live broadcast search intent, wherein the live broadcast index data includes merged static data streams and dynamic data streams corresponding to the target live broadcast room identifier.

[0009] In one embodiment, if the search intent includes a live broadcast search intent, retrieving a target live broadcast room identifier from live broadcast index data corresponding to the live broadcast search intent, comprises:

[0010] if the search intent includes a live broadcast search intent, matching the input text with a word segmentation unit in an index data block constructed according to the live broadcast index data to obtain a target word segmentation unit, wherein the word segmentation unit is obtained by word segmentation processing according to the live broadcast index data;

[0011] using a live broadcast room identifier corresponding to the target word segmentation unit as the target live broadcast room identifier.

[0012] In one of the embodiments, if the search intention includes a live broadcast search intention, the target live broadcast room identifier is retrieved from live broadcast index data corresponding to the live broadcast search intention, including:

[0013] If the search intention includes a live broadcast search intention, the original live broadcast room identifier is retrieved from the live broadcast index data according to the input text;

[0014] Obtain the live broadcast room association information corresponding to the original live broadcast room identifier;

[0015] Filter the live broadcast room association information, and obtain the original live broadcast room identifier corresponding to the filtered live broadcast room association information as the target live broadcast room identifier.

[0016] In one of the embodiments, after the target live broadcast room identifier is retrieved from the live broadcast index data corresponding to the live broadcast search intention if the search intention includes a live broadcast search intention, it further includes:

[0017] When the number of target live broadcast room identifiers is more than one, a pre-configured sorting model is used to sort the target live broadcast room identifiers to obtain the sorted target live broadcast room identifiers.

[0018] In one of the embodiments, the generation method of the live broadcast index data includes:

[0019] Receive multiple data streams, the data stream carries a live broadcast room identifier, and the data stream includes static data stream and dynamic data stream;

[0020] Merge the static data stream and the dynamic data stream corresponding to the same live broadcast room identifier to obtain the live broadcast index data corresponding to the live broadcast room identifier.

[0021] In one of the embodiments, after receiving the multiple data streams, it further includes:

[0022] Clean the static data stream and the dynamic data stream to obtain the cleaned static data stream and the cleaned dynamic data stream;

[0023] In this embodiment, the static data stream and the dynamic data stream corresponding to the same live broadcast room identifier are merged to obtain the live broadcast index data corresponding to the live broadcast room identifier, including:

[0024] Merge the cleaned static data stream and the cleaned dynamic data stream corresponding to the same live broadcast room identifier to obtain the live broadcast index data corresponding to the live broadcast room identifier.

[0025] In one of the embodiments, the static data stream and the dynamic data stream corresponding to the same live broadcast room identifier are merged to obtain the live broadcast index data corresponding to the live broadcast room identifier, including:

[0026] acquire static data stream in the received multi-path data stream when the preset update time arrives;

[0027] merge the received dynamic data stream and the static data stream when the preset update time arrives, to obtain live index data corresponding to the live room identifier.

[0028] In one of the embodiments, after the static data stream and the dynamic data stream corresponding to the same live room identifier are merged to obtain the live index data corresponding to the live room identifier, the method further includes:

[0029] acquire the number of the plurality of index servers;

[0030] convert the live room identifier to obtain a corresponding identifier value;

[0031] perform a modulo operation on the identifier value according to the number of the plurality of index servers, to obtain a data distribution identifier;

[0032] send the live index data to an index server corresponding to the data distribution identifier.

[0033] In one of the embodiments, after the target live room identifier is retrieved from the live index data corresponding to the live search intent if the search intent includes the live search intent, the method further includes:

[0034] generate a live room list according to the target live room identifier;

[0035] send the live room list to the audience client, the live room list being used to instruct the audience client to display live room related information corresponding to the target live room identifier in a page.

[0036] According to a second aspect of the embodiments of the present disclosure, a live room searching apparatus is provided, including:

[0037] an acquisition module configured to acquire input text;

[0038] an intent analysis module configured to perform intent analysis on the input text to obtain at least one search intent;

[0039] a retrieval module configured to retrieve a target live room identifier from live index data corresponding to a live search intent if the search intent includes the live search intent, the live index data including merged static data stream and dynamic data stream corresponding to the target live room identifier.

[0040] In one of the embodiments, the retrieval module is configured to perform the following: when the search intention includes a live search intention, matching the input text with word segmentation units in an index data block constructed according to live index data, obtaining a target word segmentation unit matched with the input text, the word segmentation unit being obtained through word segmentation processing according to the live index data;

[0041] taking a live room identifier corresponding to the target word segmentation unit as a target live room identifier.

[0042] In one of the embodiments, the retrieval module includes:

[0043] The retrieval unit is configured to perform the following: if the search intention includes a live search intention, obtaining a raw live room identifier from the live index data according to the input text;

[0044] The associated information obtaining unit is configured to perform the following: obtaining live room associated information corresponding to the raw live room identifier;

[0045] The filtering unit is configured to perform the following: filtering the live room associated information, and obtaining a raw live room identifier corresponding to filtered live room associated information as a target live room identifier.

[0046] In one of the embodiments, the apparatus further includes:

[0047] The sorting module is configured to perform the following: when the number of target live room identifiers is more than one, sorting the target live room identifiers using a pre-configured sorting model to obtain sorted target live room identifiers.

[0048] In one of the embodiments, the apparatus further includes:

[0049] The receiving module is configured to perform the following: receiving multiple data streams, the data streams carrying live room identifiers, the data streams including static data streams and dynamic data streams;

[0050] The merging module is configured to perform the following: merging static data streams and dynamic data streams corresponding to the same live room identifier to obtain live index data corresponding to the live room identifier.

[0051] In one of the embodiments, the apparatus further includes:

[0052] The data cleaning module is configured to perform the following: performing cleaning processing on the static data streams and the dynamic data streams to obtain cleaned static data streams and cleaned dynamic data streams;

[0053] The merging module is configured to perform the following: merging the cleaned static data streams and the cleaned dynamic data streams corresponding to the same live room identifier to obtain live index data corresponding to the live room identifier.

[0054] In one of the embodiments, the merging module comprises:

[0055] The static data stream determination unit is configured to perform determination of a static data stream in the multiple data streams received at the preset update time.

[0056] The merging unit is configured to perform merging of the received dynamic data stream and the static data stream at the preset update time to obtain the live index data corresponding to the live room identifier.

[0057] In one of the embodiments, the obtaining module is further configured to perform obtaining of the number of the multiple index servers.

[0058] The apparatus further comprises a conversion module configured to perform conversion processing of the live room identifier to obtain a corresponding identifier value.

[0059] The data distribution identifier determination module is configured to perform modulo processing of the identifier value according to the number of the multiple index servers to obtain the data distribution identifier.

[0060] The sending module is configured to perform sending of the live index data to the index server corresponding to the data distribution identifier.

[0061] In one of the embodiments, the apparatus further comprises:

[0062] The live room list generation module is configured to perform generation of a live room list according to the target live room identifier.

[0063] The sending module is configured to perform sending of the live room list to the audience client, and the live room list is used to instruct the audience client to display live room related information corresponding to the target live room identifier in a page.

[0064] According to a third aspect of the embodiments of the present disclosure, a server is provided, comprising:

[0065] A processor;

[0066] A memory for storing instructions executable by the processor;

[0067] The processor is configured to execute the instructions to implement the live room searching method in any one of the embodiments of the first aspect.

[0068] According to a fourth aspect of the embodiments of the present disclosure, a storage medium is provided, when instructions in the storage medium are executed by a processor of a server, the server is enabled to perform the live room searching method in any one of the embodiments of the first aspect.

[0069] According to a fifth aspect of the embodiments of the present disclosure, a computer program product is provided, the program product comprising a computer program stored in a readable storage medium, at least one processor of a device reading and executing the computer program from the readable storage medium, so that the device executes the live room searching method described in any one of the embodiments of the first aspect.

[0070] The technical solutions provided by the embodiments of the present disclosure bring at least the following beneficial effects:

[0071] By introducing the multi-path search strategy architecture, the input text of the client is analyzed for intent to obtain at least one search intent; for each search intent, the index data corresponding to each search intent is retrieved, so that the search of the live room is based on user preferences, spatiotemporal characteristics, context, interaction, and other multi-modal information, thereby improving the accuracy of live room search; and because the multi-path search strategy architecture is adopted, the algorithm can support concurrent strategy updating, thereby improving the updating efficiency; for live search intent, live index data is constructed according to the dynamic data stream and the static data stream corresponding to the live room identifier, and the real-time dynamic change characteristics of the live room data are utilized, which can effectively improve the accuracy of live room search, thereby greatly improving the search volume of users.

[0072] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and cannot limit the present disclosure. BRIEF DESCRIPTION OF DRAWINGS

[0073] The accompanying drawings, which are incorporated into and form part of the specification, illustrate embodiments consistent with the present disclosure and, together with the specification, serve to explain the principles of the present disclosure, and do not constitute an undue limitation on the present disclosure.

[0074] Figure 1 is an application environment diagram of a live room searching method according to an exemplary embodiment.

[0075] Figure 2 is a flowchart of a live room searching method according to an exemplary embodiment.

[0076] Figure 3 is a flowchart of a step of obtaining a target live room identifier according to an exemplary embodiment.

[0077] Figure 4 is a flowchart of a step of generating live index data according to an exemplary embodiment.

[0078] Figure 5 is a flowchart of a step of sending live index data according to an exemplary embodiment.

[0079] Figure 6 is a flow chart of a live room search method according to an example embodiment.

[0080] Figure 7 is an index structure diagram of a live room search according to an example embodiment.

[0081] Figure 8 is a schematic diagram of logical sharding of live index data according to an example embodiment.

[0082] Figure 9 is a block diagram of a live room search device according to an example embodiment.

[0083] Figure 10 is an internal structure diagram of a server according to an example embodiment. DETAILED DESCRIPTION

[0084] In order to make the ordinary person in the art better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be described clearly and completely below in conjunction with the drawings.

[0085] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The implementation described in the following example embodiments does not represent all implementations consistent with the present disclosure. Rather, they are only examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0086] The live room search method provided by the present disclosure can be applied to, for example, Figure 1The application environment shown. Among them, several audience terminals 110 and servers 120 communicate through the network, several anchors 130 and servers 120 communicate through the network. The application program installed in the audience terminal 110 can be used to watch live, and the audience can search the live room through the application program. The application program installed in the anchor terminal 130 can be used to live. The application program installed in the audience terminal 110 for watching live and the application program installed in the anchor terminal 130 for live can be the same application program. During the process of live in the anchor terminal 130, the server 120 receives the multi-channel data stream sent by the anchor terminal 130 carrying the live room identifier. The multi-channel data stream includes static data stream and dynamic data stream. The server 120 merges the static data stream and the dynamic data stream corresponding to the live room identifier, and generates the live index data corresponding to the live room identifier. When the server 120 receives the input text sent by the audience terminal 110 for live search, the input text is analyzed to obtain at least one search intent; if the search intent includes the live search intent, the target live room identifier is retrieved from the live index data corresponding to the live search intent. Among them, the audience terminal 110 can be but not limited to various personal computers, notebook computers, smart phones, tablet computers, the server 120 can be realized by independent server or multiple servers composed of server cluster, the anchor terminal 130 can be but not limited to various personal computers, notebook computers, smart phones, tablet computers.

[0087] Figure 2 The flow chart of a live room search method according to an exemplary embodiment is shown, as Figure 2 The live room search method is used in the server 120, including the following steps.

[0088] In step S210, the input text is obtained.

[0089] In step S220, the input text is analyzed to obtain at least one search intent.

[0090] Specifically, when the audience needs to search for a live room, the audience inputs text through a search entry of the audience end, for example, inputs "food live room". The audience end sends a search request for the live room to the server, and the search request carries the input text. The server responds to the received search request, performs intent analysis on the input text carried in the search request, and obtains at least one search intent. The intent analysis on the input text can adopt a classification method based on a rule template or a classification method based on machine learning. For example, if the classification method based on machine learning is adopted, different search intent categories can be defined in advance according to the characteristics of the live room, such as a live label search intent, a game live search intent, and a live search intent. Then, the common words corresponding to each intent category are defined. For the obtained input text, the probability of each intent category is calculated by the classification model based on machine learning according to the pre-defined common words, and the search intent corresponding to the intent category with a probability greater than a threshold is obtained.

[0091] In step S230, if the search intent includes a live search intent, the target live room identifier is retrieved from the live index data corresponding to the live search intent, and the live index data includes the merged static data stream and dynamic data stream corresponding to the target live room identifier.

[0092] The search intent includes at least a live search intent. The live search intent refers to the intent of searching from a live room in a live broadcast according to the input text. The target live room identifier refers to the retrieved live room identifier matching the input text. The live room identifier is used to uniquely distinguish different live rooms, for example, can be a host ID, a host name, etc. The static data stream refers to data that does not change in a period of time during the live broadcast, for example, host associated information of the live room, a live room title, ACU (Average concurrent users) of the live room, etc. The dynamic data stream refers to data that changes in real time during the live broadcast, for example, a voice data stream, a comment data stream, a live video data stream, etc.

[0093] Specifically, a multi-path search strategy architecture is constructed in advance for live room search. Each search strategy corresponds to a search intent; each search intent corresponds to specific index data. For the live search intent, the live index data corresponding thereto is generated in real time according to the multi-path data stream sent by the host end in the live broadcast. If the at least one search intent obtained by performing intent analysis on the input text includes the live search intent, the target live room identifier matching the input text is retrieved from the live index data corresponding to the live search intent.

[0094] In the search method of the live room, a multi-path search strategy architecture is introduced to analyze the input text of the client to obtain at least one search intention; for each search intention, the index data corresponding to each search intention is retrieved to make the search of the live room based on multi-modal information such as user preference, space-time characteristics, context, and interaction, thereby improving the accuracy of the live room search; and because the multi-path search strategy architecture is adopted, the algorithm can concurrently update the strategy, thereby improving the update efficiency; for the live search intention, the live index data is constructed according to the dynamic data stream and the static data stream corresponding to the live room identifier, and the real-time dynamic change of the live room data can effectively improve the accuracy of the live room search, thereby greatly improving the search volume of the user.

[0095] In an example embodiment, in step S230, if the search intention includes a live search intention, the target live room identifier is retrieved from the live index data corresponding to the live search intention, including: if the search intention includes a live search intention, the input text is matched with the token units in the index data block constructed according to the live index data, and the target token unit matched with the input text is obtained; and the live room identifier corresponding to the target token unit is taken as the target live room identifier.

[0096] Specifically, in order to improve the search efficiency of the live room search, in the present embodiment, the live index data is logically fragmented. The logical fragmentation refers to dividing the live index data into multiple smaller index data blocks. First, after generating the live index data according to the static data stream and the dynamic data stream sent by the anchor end, the live index data is processed by a token algorithm and lexical analysis to obtain multiple token units. The token algorithm can adopt a token method based on string matching, a semantic token method, a statistical token method, etc. Then, a preset number of token units form an index data block (chunk). The preset number can be 1024. Finally, the multiple index data blocks form a fragmented live index library. After receiving the input text, the input text is matched with the token units in the index data block to obtain the target token unit matched with the input text. The live room identifier corresponding to the target token unit is found as the target live room identifier.

[0097] In the present embodiment, the live index data is logically fragmented, so that the input text is matched with the token units in the index data block, and compared with the unfragmented live index data, fewer token units can be traversed, thereby speeding up the retrieval response time, faster obtaining the search result, and efficiently improving the search efficiency.

[0098] In an example embodiment, as Figure 3As shown, in step S230, if the search intention includes a live search intention, the target live room identifier is retrieved from the live index data corresponding to the live search intention. The retrieval can be achieved through the following steps.

[0099] In step S231, if the search intention includes a live search intention, the original live room identifier is retrieved from the live index data according to the input text.

[0100] In step S232, the live room associated information corresponding to the original live room identifier is obtained.

[0101] In step S233, the live room associated information is filtered, and the original live room identifier corresponding to the filtered live room associated information is obtained as the target live room identifier.

[0102] The original live room identifier refers to the live room identifier retrieved from the live index data without processing. Specifically, after receiving the input text, the live room identifier retrieved from the live index data according to the input text is taken as the original live room identifier. The live room corresponding to the original live room identifier may include high-risk live rooms, such as illegal live rooms, sensitive topic live rooms, etc. Therefore, the original live room identifier needs to be filtered, and the filtered original live room identifier is taken as the target live room identifier.

[0103] The filtering of the original live room identifier is not limited to the pre-deployed risk control strategy, blacklist filtering, and manual intervention. For example, the risk control strategy can be based on pre-deployed multi-dimensional rules, such as audience scores corresponding to the live room identifier, historical behavior data corresponding to the live room identifier, anchor characteristics corresponding to the live room identifier, etc. If the live room corresponding to the original live room identifier is determined to be a high-risk live room by the risk control strategy, the original live room identifier is filtered.

[0104] In this embodiment, by using multiple data filtering methods and filtering the retrieved original live room identifier based on multiple dimensions, the amount of live room information obtained by the audience can be reduced, and the efficiency of the audience in finding interested resources can be improved. At the same time, by filtering the original live room identifier and selecting the live room identifier meeting the preset requirements for pushing, the reliability of the live room information obtained by the audience can be ensured.

[0105] In an exemplary embodiment, in step S230, after retrieving the target live room identifier from the live index data corresponding to the live search intention if the search intention includes a live search intention, the method further includes: when the number of target live room identifiers is multiple, using a pre-configured sorting model to sort the target live room identifiers to obtain sorted target live room identifiers.

[0106] Specifically, since the live application usually sends the retrieved target live room identifier corresponding live room related information to the audience end in the form of a list. Therefore, in order to enable the audience to quickly obtain the live room related information that matches the input text, after obtaining the target live room identifier, the target live room identifier is sorted by a pre-configured sorting model to obtain a sorted target live room identifier. In this embodiment, since the search of the live room is implemented by a multi-path search strategy architecture, the priority of each search intent can be pre-configured. If multiple search intents are obtained by performing intent analysis on the input text, the target live room identifier is sorted according to the priority of each search intent. For the target live room identifier corresponding to the same search intent, the target live room identifier can be sorted according to the live room related information and search logs and the like by a pre-deployed sorting model. The live room related information is not limited to including the number of online users, audience retention time, the number of comments, live content (such as hot topic live), and the like. The sorting model calculates the sorting value corresponding to each target live room identifier, and sorts the target live room identifiers corresponding to the same search intent.

[0107] In this embodiment, by sorting the retrieved target live room identifier, the audience can quickly obtain the live room related information that matches the input text, thereby reducing the time of the audience to select the live room to be watched, improving the audience's viewing time in the live room, and further helping to improve the audience's retention rate.

[0108] In an example embodiment, as shown in Figure 4 the generation of live index data is described, including the following steps.

[0109] In step S410, a plurality of data streams are received, the data streams carry live room identifiers, and the data streams include static data streams and dynamic data streams.

[0110] In step S420, the static data stream and the dynamic data stream corresponding to the same live room identifier are merged to obtain live index data corresponding to the live room identifier.

[0111] Specifically, the search of the live room has the characteristics of real-time and state. The state refers to the state of the live room including the on-air state and the off-air state. The search of the live room is for the live room in the on-air state, and the data stream of the live room is updated in real time with the progress of the live broadcast. When the server receives the multiple data streams carrying the live room identifier, it is judged whether the current live room state corresponding to the live room identifier is the on-air state; if yes, the multiple data streams are received; otherwise, the multiple data streams are discarded. Each data stream received by the server carries a corresponding data type, including dynamic data and static data. The data stream with the data type of dynamic data is regarded as a dynamic data stream, and the data stream with the data type of static data is regarded as a static data stream. The static data stream and the dynamic data stream corresponding to the live room identifier are merged to form the live index data corresponding to the live room identifier.

[0112] In the embodiment, the live index data is obtained by merging the dynamic data stream and the static data stream corresponding to the live room identifier, and the state characteristics and the real-time dynamic change characteristics of the live room data can effectively improve the accuracy of the search of the live room, thereby greatly improving the search amount of the user.

[0113] In an exemplary embodiment, after receiving the multiple data streams in step S410, the static data stream and the dynamic data stream are further subjected to cleaning processing to obtain cleaned static data stream and cleaned dynamic data stream. In the embodiment, the static data stream and the dynamic data stream corresponding to the same live room identifier are merged to obtain the live index data corresponding to the live room identifier, including: merging the cleaned static data stream and the cleaned dynamic data stream corresponding to the same live room identifier to obtain the live index data corresponding to the live room identifier.

[0114] Specifically, since there are often a large amount of noise data in the live data received by the server, for example, sensitive words, meaningless words (such as, ah), etc., in order to improve the accuracy of the live index data, after receiving the multiple data streams, the static data stream and the dynamic data stream can be subjected to cleaning processing to obtain cleaned static data stream and cleaned dynamic data stream. The cleaning processing can be performed by using the reference factors of the pre-defined data cleaning, for example, the anchor characteristic data, the title of the live room, the number of online people in the live room, the score of the live room, the preset dictionary, etc. The live room data meeting the preset requirements is cleaned from the received multiple data streams according to the reference factors.

[0115] In the embodiment, after receiving the multiple data streams, noise data in the multiple data streams is cleaned, on one hand, after the data cleaning, the reduction of the data quantity helps to improve the generation efficiency of the live index data, and the operation pressure of the server is reduced; on the other hand, through the cleaning of the data, the quality of the index data is improved, and the accuracy of the live index data is improved.

[0116] In an example embodiment, in step S420, the static data stream and the dynamic data stream corresponding to the same live room identifier are merged to obtain the live index data corresponding to the live room identifier, which can specifically include: obtaining the static data stream in the multiple data streams received when the preset update time arrives; merging the received dynamic data stream and the static data stream when the preset update time arrives to obtain the live index data corresponding to the live room identifier.

[0117] Specifically, in order to improve the efficiency of generating the live index data, different update strategies are adopted for the dynamic data stream and the static data stream in the embodiment. Since the static data stream is a data stream that will not be updated within a certain time period, the corresponding update time is configured in advance for the static data, and the update time can be a periodic time. After receiving the multiple data streams, the server judges whether the current time has arrived at the update time of the static data stream, if not, the static data stream remains unchanged, and the dynamic data stream currently received and the static data stream stored are merged to obtain the live index data. If it is judged that the current time has arrived at the update time of the static data stream, the static data stream currently received and the dynamic data stream are merged to obtain the live index data.

[0118] In the embodiment, by using the time update characteristics of the dynamic data stream and the static data stream, the corresponding update strategies are configured for the dynamic data stream and the static data stream respectively, which can improve the processing efficiency of the data and reduce the operation pressure of the server.

[0119] In an example embodiment, as shown in Figure 5 After the static data stream and the dynamic data stream corresponding to the same live room identifier are merged to obtain the live index data corresponding to the live room identifier, the following steps are further included.

[0120] In step S510, the number of multiple index servers is obtained.

[0121] In step S520, the live room identifier is converted to obtain a corresponding identifier value.

[0122] In step S530, the identifier value is processed by modulo according to the number of multiple index servers to obtain a data distribution identifier.

[0123] In step S540, the live index data is sent to an index server corresponding to the data distribution identifier.

[0124] Specifically, after obtaining the live index data, the live index data is subjected to physical sharding processing. The physical sharding processing refers to distributing the live index data on multiple separate index server nodes. The physical sharding processing can adopt a hash-based sharding manner, a data range-based sharding manner, etc. In this embodiment, the hash-based sharding manner is adopted. The hash-based sharding manner in this embodiment is described below. First, the number of index servers is obtained. The live room identifier is subjected to conversion processing by using a hash algorithm, and the obtained hash value is taken as an identifier value. Then, the identifier value is subjected to modulo processing according to the number of index servers, and a data distribution identifier is obtained. The data distribution identifier is used to uniquely distinguish the index servers, and the correspondence between the data distribution identifier and the index server is set in advance. Finally, the index server corresponding to the live room identifier is found from the correspondence between the data distribution identifier and the index server according to the obtained data distribution identifier, and the live index data corresponding to the live room identifier is sent to the index server.

[0125] In this embodiment, by uniformly distributing the live index data corresponding to multiple live room identifiers into multiple index servers, when part of the index servers fail, the normal live room search will not be affected, thereby ensuring the reliability of the live room search. By distributing the live index data into multiple index servers, the running pressure of each index server can be reduced, and the data retrieval time can be accelerated, thereby improving the search efficiency of the live room.

[0126] In an example embodiment, after the target live room identifier is obtained from the live index data corresponding to the live search intent, if the search intent includes the live search intent, the method further includes: generating a live room list according to the target live room identifier; and sending the live room list to the audience client, the live room list being used to instruct the audience client to display live room related information corresponding to the target live room identifier in a page.

[0127] The live room related information includes information displayed on the audience client interface and used to distinguish different live rooms, for example, can be a host avatar of a live room, a live picture of a live room, etc. Specifically, after obtaining the target live room identifier, the server obtains live room related information corresponding to the live room identifier. The server combines the obtained target live room identifier and the corresponding live room related information into a live room list, and sends the live room list to the audience client, so that the audience client can display the live room related information corresponding to each live room in the live recommendation list in a preset position of the current interface.

[0128] Figure 6is a flow chart of a search method of a live room according to an exemplary embodiment; Figure 7 is an index structure diagram of a live room search according to an exemplary embodiment. The following will be described in combination with Figure 6 and Figure 7 The search method of the live room is specifically described, including the following steps.

[0129] In step S601, multiple data streams are received, the data streams carry live room identifiers, and the data streams include static data streams and dynamic data streams.

[0130] Specifically, the received multiple data streams are data streams determined by the server according to the live room identifier to be in a live state. Each data stream is pre-configured with a corresponding data type, including dynamic data and static data. The data stream with a static data type is a static data stream; the data stream with a dynamic data type is a dynamic data stream. As shown in Figure 7 the live room video data stream, the game live room video data stream, the ASR (Automatic Speech Recognition) data stream, and the comment data stream are dynamic data streams; the anchor related information, the live title, and other information are static data streams.

[0131] In step S602, the dynamic data streams and the static data streams to be merged are determined.

[0132] Specifically, according to the update strategy pre-configured for the dynamic data and the static data, the corresponding static data stream and the dynamic data stream are obtained. For the dynamic data stream, a real-time update strategy is adopted; for the static data stream, a periodic update strategy is adopted. That is, after receiving the multiple data streams, it is determined whether the current time has reached the update time of the static data stream. If it has not reached, the static data stream remains unchanged. If it is determined that the current time has reached the update time of the static data stream, the stored static data stream is updated to the currently received static data stream.

[0133] In step S603, the static data stream and the dynamic data stream are cleaned to obtain a cleaned static data stream and a cleaned dynamic data stream.

[0134] In step S604, the cleaned static data stream and the cleaned dynamic data stream corresponding to the same live room identifier are merged to obtain live index data corresponding to the live room identifier.

[0135] In step S605, the live index data is sent to an index server.

[0136] Specifically, the live index data is physically sharded and sent to multiple separate index server nodes. The physical sharding can adopt a hash-based sharding manner, and details can be referred to the description of the above embodiments, which will not be described here.

[0137] In step S606, the live index data in the index server is logically sharded.

[0138] Specifically, the live index data is tokenized to obtain a plurality of token units; and each 1024 token units form an index data block. Figure 8 An exemplary diagram of obtaining a plurality of index data blocks from a plurality of token units is shown.

[0139] In step S607, the input text is obtained.

[0140] In step S608, the input text is analyzed to obtain at least one search intent.

[0141] In step S609, if the search intent includes a live search intent, the original live room identifier is retrieved from the live index data corresponding to the live search intent.

[0142] As shown in Figure 7 The search intent includes an artificial intervention top search intent, a member user query search intent, a game live search intent, a live label search intent, a popular live search intent, a live recommendation intent, and a live search intent. Among them, the game live search intent, the live label search intent, the popular live search intent, and the live search intent each correspond to independent index data. After obtaining at least one search intent, retrieval is performed from the index data corresponding to the search intent.

[0143] If the search intent includes a live search intent, the original live room identifier is retrieved from the live index data generated according to the above steps S601-S609. Specifically, the input text is matched with the token units in the data block generated according to the live index data to obtain target token units matched with the input text; and the live room identifier corresponding to the target token units is taken as the original live room identifier.

[0144] Further, if the search intent obtained by intent analysis includes other live search intents, the original live room identifier corresponding to the other live search intents can be obtained by the following way:

[0145] For the artificial intervention top search intent, artificial operation is completed through an artificial operation platform.

[0146] For the member user query search intent, the member live room identifier corresponding to the member live room can be queried according to the member live room identifier of the member live room in a state of going on air. According to the live room related information corresponding to the queried live room identifier, such as a live room title, anchor characteristics, and the like, the original live room identifier matching the input text is found.

[0147] For the game live search intent, the label data corresponding to the game live room identifier can be obtained, and the game live index data is constructed according to the label data corresponding to the game live room identifier. The input text is matched with the game live index data to obtain the original live room identifier corresponding to the game live search intent. The label data corresponding to the game live room identifier can be obtained by analyzing the data stream of the game live room through a pre-configured label model.

[0148] For the live label search intent, the label data corresponding to the live room identifier can be obtained, and the live label index data is constructed according to the label data corresponding to the live room identifier. The input text is matched with the live label index data to obtain the original live room identifier corresponding to the live label search intent. The label data corresponding to the live room identifier can be obtained by analyzing the data stream of the live room through a pre-configured label model.

[0149] For the popular live search intent, the anchor portrait characteristics corresponding to the current popular live room identifier can be obtained; and the anchor portrait characteristics index data corresponding to the popular live search intent is constructed through the anchor portrait characteristics. The input text is matched with the anchor portrait characteristics index data to obtain the original live room identifier corresponding to the popular live search intent. The popular live identifier can be obtained according to the live room related information corresponding to the live room identifier in a state of going on air, and the live room related information includes the number of audience, the number of audience comments, and the like.

[0150] In step S610, the live room state of the original live room identifier is obtained, and the original live room identifier in a state of going on air is filtered.

[0151] In step S611, the original live room identifier in a state of going on air is filtered, and the filtered original live room identifier is obtained as a target live room identifier. The filtering of the original live room identifier is not limited to the pre-deployed risk control strategy, blacklist filtering, and manual intervention.

[0152] In step S612, a pre-configured sorting model is used to sort the target live room identifier to obtain a sorted target live room identifier.

[0153] In step S613, a live room list is generated according to the target live room identifier, and the live room list is sent to the audience client.

[0154] It should be understood that, although Figures 1-8The steps in the flowchart are displayed in sequence according to the direction of the arrows, but the steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, the execution of the steps is not strictly limited in sequence, and the steps can be executed in other orders. Moreover, Figures 1-8 At least part of the steps in the flowchart can include multiple steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of the steps or stages is not necessarily sequential, but can be alternately executed with other steps or steps or stages in other steps.

[0155] Figure 10 FIG. 9 is a block diagram of a live room search device 900 according to an example embodiment. Referring to FIG. 9, Figure 9 The device includes an acquisition module 901, an intent analysis module 902, and a retrieval module 903.

[0156] The acquisition module 901 is configured to perform acquisition of input text; the intent analysis module 902 is configured to perform intent analysis on the input text to obtain at least one search intent; and the retrieval module 903 is configured to perform, if the search intent includes a live search intent, retrieval of a target live room identifier from live index data corresponding to the live search intent, the live index data including merged static data streams and dynamic data streams corresponding to the target live room identifier.

[0157] In an example embodiment, the retrieval module 903 is configured to perform, when the search intent includes a live search intent, matching of the input text with token units in index data blocks constructed according to the live index data, obtaining target token units matched with the input text, the token units being obtained by tokenization processing according to the live index data; and obtaining live room identifiers corresponding to the target token units as the target live room identifier.

[0158] In an example embodiment, the retrieval module 903 includes a retrieval unit configured to perform, if the search intent includes a live search intent, retrieval of an original live room identifier from the live index data according to the input text; an associated information acquisition unit configured to perform acquisition of live room associated information corresponding to the original live room identifier; and a filtering unit configured to perform filtering of the live room associated information, obtaining an original live room identifier corresponding to the filtered live room associated information as the target live room identifier.

[0159] In an example embodiment, the device further includes a sorting module configured to perform, when the number of target live room identifiers is multiple, sorting of the target live room identifiers using a preconfigured sorting model to obtain sorted target live room identifiers.

[0160] In an example embodiment, the apparatus further comprises: a receiving module configured to perform receiving multiple data streams, the data streams carrying live room identifiers, the data streams comprising static data streams and dynamic data streams; and a merging module configured to perform merging the static data streams and the dynamic data streams corresponding to a same live room identifier to obtain live index data corresponding to the live room identifier.

[0161] In an example embodiment, the apparatus further comprises: a data cleaning module configured to perform cleaning the static data streams and the dynamic data streams to obtain cleaned static data streams and cleaned dynamic data streams; and a merging module configured to perform merging the cleaned static data streams and the cleaned dynamic data streams corresponding to a same live room identifier to obtain live index data corresponding to the live room identifier.

[0162] In an example embodiment, the merging module comprises: a static data stream determining unit configured to perform obtaining static data streams in the multiple data streams received when a preset update time arrives; and a merging unit configured to perform merging the received dynamic data streams and the static data streams when the preset update time arrives to obtain live index data corresponding to the live room identifier.

[0163] In an example embodiment, the obtaining module 901 is further configured to perform obtaining a number of index servers; the apparatus further comprises: a conversion module configured to perform conversion processing on the live room identifier to obtain a corresponding identifier value; a data distribution identifier determining module configured to perform modulo processing on the identifier value according to the number of index servers to obtain a data distribution identifier; and a sending module configured to perform sending the live index data to an index server corresponding to the data distribution identifier.

[0164] In an example embodiment, the apparatus further comprises: a live room list generating module configured to perform generating a live room list according to a target live room identifier; and a sending module configured to perform sending the live room list to a viewer client, the live room list being used to instruct the viewer client to display live room related information corresponding to the target live room identifier in a page.

[0165] As to the apparatus in the above embodiments, the specific manners in which various modules perform operations have been described in detail in the embodiments of the method, and will not be described in detail here.

[0166] Figure 10 is a block diagram of an apparatus 1000 for live room search according to an example embodiment. For example, the apparatus 1000 can be a server. Refer to Figure 10The device 1000 also includes a processing component 1020 configured to execute instructions and manipulate data, and a memory 1022, which is accessed by the processing component 1020 and further includes both volatile and nonvolatile memory. For example, the memory 1022 can include volatile memory (e.g., random access memory (RAM)) comprising memory locations that are dynamic in nature, meaning that data stored at the memory locations is lost when the computer is turned off. The memory 1022 can also include nonvolatile memory (e.g., read-only memory (ROM), EPROM, EEPROM, hard disk, floppy disk, etc.) that retains data stored at the memory locations even when power is off. The memory 1022 can also comprise other similar structural elements known to those with ordinary skill in the art. The processing component 1020 can comprise one or more processors, microprocessors, controllers, or other processing means to manipulate data stored on the device 1000. The processing component 1020 can be configured to execute the instructions stored in the memory 1022 to implement the methods described above.

[0167] The device 1000 can also include a power supply component 1024 configured to provide power to the device 1000, a network interface 1026 configured to connect the device 1000 to a network, and an input / output (I / O) interface 1028. The device 1000 can operate based on an operating system stored in the memory 1022, such as Windows Server, Mac OSX, Unix, Linux, FreeBSD, or the like.

[0168] In exemplary embodiments, a storage medium including instructions, such as the memory 1022, is also provided, which can be executed by the processor of the device 1000 to complete the methods described above. The storage medium can be a non-transitory computer-readable storage medium, such as ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disc, and optical data storage device, etc.

[0169] Other embodiments of the present disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the features disclosed herein. It is intended that the present disclosure cover any and all variations of the disclosure including those variations that can be incorporated into other forms of the disclosure. It is intended that the specification and examples be considered exemplary only, with the true scope and spirit of the disclosure indicated by the following claims.

[0170] It is to be understood that the present disclosure is not limited to the precise construction herein described and as shown in the attached drawings, and that various modifications and changes can be effected therein by those skilled in the art without departing from the scope of the application. The scope of the application should be determined by the following claims.

Claims

1. A method for searching a live room, characterized in that, The method comprises: acquiring input text; performing intent analysis on the input text by using a multi-path search strategy architecture to obtain at least one search intent, wherein each search path corresponds to a search intent, each search intent corresponds to specific index data, and for each search intent, the search of a live broadcast room is based on user preferences, spatiotemporal characteristics, context, and multi-modal information of interaction by retrieving the index data corresponding to each search intent; if the search intent includes a live broadcast search intent, retrieving a target live broadcast room identifier from live broadcast index data corresponding to the live broadcast search intent, wherein the live broadcast index data includes merged static data streams and dynamic data streams corresponding to the target live broadcast room identifier, the static data streams are obtained by using a periodic update strategy, and the dynamic data streams are obtained by using a real-time update strategy; if the search intent includes multiple search intents, retrieving target live broadcast room identifiers by retrieving index data corresponding to each search intent, and sorting the target live broadcast room identifiers according to the priorities of the search intents.

2. The live room searching method of claim 1, wherein, If the search intent includes a live broadcast search intent, the target live broadcast room identifier is retrieved from live broadcast index data corresponding to the live broadcast search intent, which includes: if the search intent includes a live broadcast search intent, matching the input text with token units in an index data block constructed according to the live broadcast index data to obtain target token units matched with the input text, wherein the token units are obtained by tokenizing the live broadcast index data; the live broadcast room identifier corresponding to the target token unit is used as the target live broadcast room identifier.

3. The live room searching method of claim 1, wherein, If the search intent includes a live broadcast search intent, the target live broadcast room identifier is retrieved from live broadcast index data corresponding to the live broadcast search intent, which includes: if the search intent includes a live broadcast search intent, retrieving an original live broadcast room identifier from the live broadcast index data according to the input text; acquiring live broadcast room association information corresponding to the original live broadcast room identifier; filtering the live broadcast room association information to obtain an original live broadcast room identifier corresponding to filtered live broadcast room association information as the target live broadcast room identifier.

4. The live room searching method of claim 1, wherein, After the target live broadcast room identifier is retrieved from the live broadcast index data corresponding to the live broadcast search intent, the method further comprises: when the number of target live broadcast room identifiers is greater than one, sorting the target live broadcast room identifiers by using a preconfigured sorting model to obtain sorted target live broadcast room identifiers.

5. The live room searching method according to any one of claims 1-4, characterized in that, The generation method of the live broadcast index data comprises: receiving multiple data streams, wherein the data streams carry live broadcast room identifiers, and the data streams include static data streams and dynamic data streams; merging the static data streams and the dynamic data streams corresponding to the same live broadcast room identifier to obtain live broadcast index data corresponding to the live broadcast room identifier.

6. The live room searching method of claim 5, wherein, After receiving the multiple data streams, the method further comprises: performing cleaning processing on the static data streams and the dynamic data streams to obtain cleaned static data streams and cleaned dynamic data streams; The merging of the static data stream corresponding to the live room identifier and the dynamic data stream to obtain live index data corresponding to the live room identifier comprises: The merging of the cleaned static data stream corresponding to the live room identifier and the cleaned dynamic data stream to obtain live index data corresponding to the live room identifier.

7. The live room searching method of claim 5, wherein, The merging of the static data stream corresponding to the live room identifier and the dynamic data stream to obtain live index data corresponding to the live room identifier comprises: Obtaining the static data stream in the multiple data streams received when the preset update time arrives; Merging the received dynamic data stream and the static data stream at the preset update time to obtain live index data corresponding to the live room identifier.

8. The live room searching method of claim 5, wherein, After the merging of the static data stream corresponding to the live room identifier and the dynamic data stream to obtain live index data corresponding to the live room identifier, the method further comprises: Obtaining the number of multiple index servers; Converting the live room identifier to obtain a corresponding identifier value; According to the number of the multiple index servers, the identifier value is processed by modulus to obtain a data distribution identifier; The live index data is sent to the index server corresponding to the data distribution identifier.

9. The live room searching method according to any one of claims 1-4, characterized in that, After the target live room identifier is retrieved from the live index data corresponding to the live search intent if the search intent includes the live search intent, the method further comprises: Generating a live room list according to the target live room identifier; Sending the live room list to the audience client, the live room list being used to instruct the audience client to display live room related information corresponding to the target live room identifier in a page.

10. A live room searching apparatus, characterized in that, Comprise: The obtaining module is configured to perform obtaining input text; The intent analysis module is configured to perform intent analysis on the input text using a multi-path search strategy architecture to obtain at least one search intent, wherein each search strategy corresponds to one search intent, and each search intent corresponds to specific index data. For each search intent, the search of the live room is based on the multi-modal information of user preferences, spatiotemporal characteristics, context, and interaction by retrieving the index data corresponding to each search intent. The retrieval module is configured to perform retrieval of a target live room identifier from live index data corresponding to a live search intent if the search intent includes the live search intent, the live index data including merged static data stream and dynamic data stream corresponding to the target live room identifier, the static data stream being obtained using a periodic update strategy, and the dynamic data stream being obtained using a real-time update strategy. If the search intent includes multiple search intents, the target live room identifier is obtained by retrieving the index data corresponding to each search intent, and the target live room identifiers are sorted according to the priority of each search intent.

11. The live room searching apparatus according to claim 10, characterized in that, The retrieval module is configured to perform the following: when the search intention comprises a live broadcast search intention, matching the input text with word segmentation units in an index data block constructed according to the live broadcast index data, obtaining a target word segmentation unit matched with the input text, the word segmentation unit being obtained through word segmentation processing according to the live broadcast index data; obtaining a live broadcast room identifier corresponding to the target word segmentation unit as the target live broadcast room identifier.

12. The live room searching apparatus according to claim 10, wherein, The retrieval module comprises: a retrieval unit configured to perform the following: if the search intention comprises a live broadcast search intention, retrieving an original live broadcast room identifier from the live broadcast index data according to the input text; an associated information acquisition unit configured to perform the following: acquiring live broadcast room associated information corresponding to the original live broadcast room identifier; a filtering unit configured to perform the following: filtering the live broadcast room associated information, and obtaining an original live broadcast room identifier corresponding to filtered live broadcast room associated information as the target live broadcast room identifier.

13. The live room searching apparatus according to claim 10, wherein, The device further comprises: a sorting module configured to perform the following: when the number of target live broadcast room identifiers is more than one, sorting the target live broadcast room identifiers using a preconfigured sorting model to obtain sorted target live broadcast room identifiers.

14. The live room searching apparatus according to any one of claims 10-13, characterized in that, The device further comprises: a receiving module configured to perform the following: receiving multiple data streams, the data streams carrying live broadcast room identifiers, the data streams comprising static data streams and dynamic data streams; a merging module configured to perform the following: merging static data streams and dynamic data streams corresponding to the same live broadcast room identifier to obtain live broadcast index data corresponding to the live broadcast room identifier.

15. The live room searching apparatus according to claim 14, characterized in that, The device further comprises: a data cleaning module configured to perform the following: performing cleaning processing on the static data streams and the dynamic data streams to obtain cleaned static data streams and cleaned dynamic data streams; the merging module is configured to perform the following: merging the cleaned static data streams and the cleaned dynamic data streams corresponding to the same live broadcast room identifier to obtain live broadcast index data corresponding to the live broadcast room identifier.

16. The live room searching apparatus according to claim 14, characterized in that, The merging module comprises: a static data stream determination unit configured to perform the following: obtaining static data streams in the multiple data streams received when a preset update time arrives; a merging unit configured to perform the following: merging the received dynamic data streams and the static data streams when the preset update time arrives to obtain live broadcast index data corresponding to the live broadcast room identifier.

17. The live room searching apparatus according to claim 14, characterized in that, The obtaining module is further configured to perform the following: obtaining the number of multiple index servers; The device further comprises: a conversion module configured to perform the following: performing conversion processing on the live broadcast room identifier to obtain a corresponding identifier value; a data distribution identifier determination module configured to perform the following: performing modulo processing on the identifier value according to the number of the multiple index servers to obtain a data distribution identifier; a sending module configured to perform the following: sending the live broadcast index data to an index server corresponding to the data distribution identifier.

18. The live room searching apparatus according to any one of claims 10-13, characterized in that, The device further comprises: a live broadcast room list generation module configured to perform the following: generating a live broadcast room list according to the target live broadcast room identifier. The sending module is configured to perform sending the live room list to a viewer client, the live room list being used to instruct the viewer client to display live room related information corresponding to the target live room identifier in a page.

19. A server, characterized by Comprise: a processor; a memory for storing instructions executable by the processor; wherein the processor is configured to execute the instructions to implement the live room searching method according to any one of claims 1 to 9.

20. A storage medium, when instructions in the storage medium are executed by a processor of a server, enable the server to perform the live room searching method according to any one of claims 1 to 9.

21. A computer program product, comprising instructions therein, characterised in that, The instructions are executed by a processor of an electronic device, enabling the electronic device to perform the live room searching method according to any one of claims 1 to 9.

Images