Method and device for responding to retrieval request through KV storage device

Abstract

The present application discloses a method and apparatus for responding to a search request through a KV storage device. The method for responding to a search request through a KV storage device provided includes: providing a search element in a search formula to a machine learning component including an artificial neural network composed of multiple layers of nodes; the machine learning component provides the value node value of one or more layers of nodes of its artificial neural network to a key generation component; the key generation component generates a key, and provides the generated key to a KV storage device, and reads the value corresponding to the key from the KV storage device.

Description

Technical Field

[0001] The present application relates to KV storage, and in particular, to a method and apparatus for responding to a retrieval request through a KV storage device. Background Art

[0002] Storage devices that support the Key-Value (also referred to as "KV") storage model provide key-based read operations (Get(Key)) and write operations (Put(Key,Value)). To perform a write operation, the host provides the key and value to the storage device to write the value to the storage device, and uses the key as the index of the written value. To perform a read operation, the host provides the key to the storage device, the storage device finds the value based on the key, and provides the value to the host. Therefore, in the KV storage system, the key is the index used to access the value, and the value is the data being accessed. Generally, the length of the key and value can be fixed or variable. Summary of the invention

[0003] According to the first aspect of the present application, a method for generating keys and values ​​of a first KV storage according to the first aspect of the present application is provided, including: generating structured information based on original data; providing a search formula to a machine learning component, and generating a key based on the output of the search formula processed by the machine learning component; wherein the search formula is used to retrieve the structured information; using the search result of the search performed according to the search formula as a value; and recording the key and the value in the KV storage device.

[0004] According to the method for generating keys and values ​​of the first KV storage according to the first aspect of the present application, there is provided a method for generating keys and values ​​of the second KV storage according to the first aspect of the present application, and the method for performing structured processing on the original data includes: adding labels to the original data and / or extracting features of the original data.

[0005] According to the method for generating keys and values ​​of the second KV storage according to the first aspect of the present application, there is provided a method for generating keys and values ​​of the third KV storage according to the first aspect of the present application, wherein labels are added to the original data to indicate one or more of the source, format, storage location, and required access permissions of the original data.

[0006] According to one of the methods for generating keys and values ​​of the first to third KV storages of the first aspect of the present application, there is provided a method for generating keys and values ​​of the fourth KV storage of the first aspect of the present application, wherein the generated structured information includes retrievable elements.

[0007] According to the method for generating keys and values ​​of the fourth KV storage according to the first aspect of the present application, there is provided a method for generating keys and values ​​of the fifth KV storage according to the first aspect of the present application, and the searchable elements include: objects contained in the video or picture, features of the objects, and one or more keywords in the file summary.

[0008] According to one of the methods for generating keys and values ​​of the first to fifth KV storages of the first aspect of the present application, there is provided a method for generating keys and values ​​of the sixth KV storage according to the first aspect of the present application, wherein the search formula includes one or more search elements.

[0009] According to one of the methods for generating keys and values ​​of the first to sixth KV storages of the first aspect of the present application, there is provided a method for generating keys and values ​​of the seventh KV storage according to the first aspect of the present application, the user's search behavior is obtained, and the search formula provided by the user is obtained from the user's search behavior.

[0010] According to one of the methods for generating keys and values ​​of the first to seventh KV storages of the first aspect of the present application, there is provided a method for generating keys and values ​​of the eighth KV storage according to the first aspect of the present application, and the retrieval formula is a retrieval command provided to the search engine or database during the retrieval process.

[0011] According to one of the methods for generating keys and values ​​of the first to eighth KV storages of the first aspect of the present application, there is provided a method for generating keys and values ​​of the ninth KV storage according to the first aspect of the present application, wherein the machine learning component includes an artificial neural network composed of multiple layers of nodes, and the artificial neural network includes an input layer, one or more internal layers and an output layer in sequence, and each layer of the artificial neural network includes multiple nodes, and each node of the input layer receives an input retrieval formula.

[0012] According to the method for generating keys and values ​​of the ninth KV storage according to the first aspect of the present application, there is provided a method for generating keys and values ​​of the tenth KV storage according to the first aspect of the present application, wherein each node of the input layer receives one of the search elements constituting the search formula.

[0013] According to the ninth or tenth KV storage key and value generation method of the first aspect of the present application, there is provided an eleventh KV storage key and value generation method according to the first aspect of the present application, and the values ​​of each node of the output layer indicate the output of the machine learning component.

[0014] According to the eleventh KV storage key and value generation method of the first aspect of the present application, there is provided a twelfth KV storage key and value generation method according to the first aspect of the present application, and the output of the machine learning component is a score for the input search formula.

[0015] According to one of the methods for generating keys and values ​​of the ninth to twelfth KV storages of the first aspect of the present application, there is provided a method for generating keys and values ​​of the thirteenth KV storage according to the first aspect of the present application, and the node values ​​of each layer of the artificial neural network are connected respectively to obtain sequences corresponding to each layer of the artificial neural network.

[0016] According to the thirteenth KV storage key and value generation method of the first aspect of the present application, there is provided a fourteenth KV storage key and value generation method according to the first aspect of the present application, and hash calculations are performed on the sequences corresponding to each layer to obtain hash values ​​corresponding to each layer.

[0017] According to the fourteenth KV storage key and value generation method of the first aspect of the present application, there is provided a fifteenth KV storage key and value generation method according to the first aspect of the present application, and the hash values ​​corresponding to each layer have the same length.

[0018] According to the fourteenth or fifteenth KV storage key and value generation method of the first aspect of the present application, there is provided a sixteenth KV storage key and value generation method according to the first aspect of the present application, connecting the hash value corresponding to one or more layers of the artificial neural network as the key provided to the KV storage device.

[0019] According to the fourteenth or fifteenth KV storage key and value generation method of the first aspect of the present application, there is provided a seventeenth KV storage key and value generation method according to the first aspect of the present application, and the connection results of the hash values ​​corresponding to one or more layers of the artificial neural network are further hashed, and the obtained results are used as the keys provided to the KV storage device.

[0020] According to the fourteenth or fifteenth KV storage key and value generation method of the first aspect of the present application, there is provided an eighteenth KV storage key and value generation method according to the first aspect of the present application, which generates keys provided to the KV storage device based on the values ​​of one or more layers of nodes of the artificial neural network.

[0021] According to one of the methods for generating keys and values ​​of the first to eighteenth KV storages of the first aspect of the present application, there is provided a method for generating keys and values ​​of the nineteenth KV storage of the first aspect of the present application, and the machine learning component generates a score for the search formula as an output based on the input search formula.

[0022] According to the nineteenth KV storage key and value generation method of the first aspect of the present application, there is provided a twentieth KV storage key and value generation method according to the first aspect of the present application, and the accuracy of the scoring is evaluated based on the quality of the search formula.

[0023] According to the method for generating keys and values ​​of the twentieth KV storage according to the first aspect of the present application, there is provided a method for generating keys and values ​​of the twenty-first KV storage according to the first aspect of the present application, wherein the machine learning component is trained to evaluate the quality of the input search formula through the output score.

[0024] According to the twenty-first KV storage key and value generation method of the first aspect of the present application, there is provided a twenty-second KV storage key and value generation method according to the first aspect of the present application, wherein the machine learning component is trained to generate similar scores for retrieval expressions with similar retrieval results.

[0025] According to the twenty-second KV storage key and value generation method of the first aspect of the present application, there is provided a twenty-third KV storage key and value generation method according to the first aspect of the present application, which uses the retrieval results of the implemented retrieval as an evaluation of the score of the retrieval formula output by the machine learning component, so that the retrieval formula machine learning component outputs similar scores for similar retrieval results.

[0026] According to one of the methods for generating keys and values ​​of the first to twenty-third KV storages of the first aspect of the present application, there is provided a method for generating keys and values ​​of the twenty-fourth KV storage of the first aspect of the present application, which uses the subsequent behavior of the user after the retrieval as an evaluation of the score of the retrieval formula output by the machine learning component, and provides the evaluation to the machine learning component.

[0027] According to the twenty-fourth KV storage key and value generation method of the first aspect of the present application, there is provided the twenty-fifth KV storage key and value generation method according to the first aspect of the present application. The user clicks on one or more search results provided by the database or search engine, which means that the search formula is better.

[0028] According to the twenty-fourth KV storage key and value generation method of the first aspect of the present application, there is provided the twenty-sixth KV storage key and value generation method according to the first aspect of the present application, wherein the user ignores all or most of the search results provided by the database or search engine, meaning that the search formula is inferior.

[0029] According to one of the methods for generating keys and values ​​of the first to twenty-sixth KV storages of the first aspect of the present application, there is provided a method for generating keys and values ​​of the twenty-seventh KV storage of the first aspect of the present application, and the machine learning component updates the weights of its internal nodes based on the evaluation to gradually generate scores that can more accurately evaluate the quality of the retrieval formula.

[0030] According to one of the methods for generating keys and values ​​of the first to twenty-seventh KV storages of the first aspect of the present application, there is provided a method for generating keys and values ​​of the twenty-eighth KV storage of the first aspect of the present application, which classifies the retrieval results of the retrieval, and uses the difference between the value corresponding to the category and the score as an evaluation of the score of the retrieval formula output by the machine learning component.

[0031] According to one of the first to twenty-eighth KV storage key and value generation methods of the first aspect of the present application, there is provided a twenty-ninth KV storage key and value generation method according to the first aspect of the present application, and the KV storage device responds to receiving a query request for a key by outputting a value corresponding to the key as a response to the query request.

[0032] According to the second aspect of the present application, a key and value generation system for the first KV storage according to the second aspect of the present application is provided, including: a structured processing module for generating structured information based on raw data; a key generation module for providing a retrieval formula to a machine learning component, and generating a key based on the output of the retrieval formula processed by the machine learning component; wherein the retrieval formula is used to retrieve the structured information; a value generation module for using the retrieval result of the retrieval implemented according to the retrieval formula as a value; and a storage module for recording the key and the value in the KV storage device.

[0033] According to the third aspect of the present application, a method for responding to a retrieval request through a KV storage device according to the first aspect of the third aspect of the present application is provided, including: providing retrieval elements in a retrieval formula to a machine learning component; the machine learning component providing the generated node value to a key generation component; the key generation component generates a key, and provides the generated key to the KV storage device, and reads a value corresponding to the key from the KV storage device.

[0034] According to the first method of responding to a retrieval request through a KV storage device according to the third aspect of the present application, a second method of responding to a retrieval request through a KV storage device according to the third aspect of the present application is provided, and the key is the key described in any one of items 16 to 18 of the first aspect.

[0035] According to the first or second method of responding to a retrieval request through a KV storage device according to the third aspect of the present application, a third method of responding to a retrieval request through a KV storage device according to the third aspect of the present application is provided, and structured information corresponding to the same key is recorded in the value.

[0036] According to the third method of responding to a retrieval request through a KV storage device according to the third aspect of the present application, a fourth method of responding to a retrieval request through a KV storage device according to the third aspect of the present application is provided, and the structured information includes the storage location of the original data for generating the structured information.

[0037] According to one of the first to fourth methods of responding to retrieval requests through a KV storage device according to the third aspect of the present application, a fifth method of responding to retrieval requests through a KV storage device according to the third aspect of the present application is provided, using structured information corresponding to the same key obtained from the KV storage device as a response to the retrieval.

[0038] According to one of the first to fifth methods of responding to a retrieval request through a KV storage device according to the third aspect of the present application, a sixth method of responding to a retrieval request through a KV storage device according to the third aspect of the present application is provided to display structured information to the user as a retrieval result.

[0039] According to one of the first to sixth methods of responding to retrieval requests through a KV storage device according to the third aspect of the present application, a seventh method of responding to retrieval requests through a KV storage device according to the third aspect of the present application is provided, and the structured information includes thumbnails of images or videos hit according to the characteristics of the user's search.

[0040] According to one of the first to seventh methods of responding to retrieval requests through a KV storage device according to the third aspect of the present application, there is provided an eighth method of responding to retrieval requests through a KV storage device according to the third aspect of the present application, wherein thumbnails are displayed to users to facilitate users to identify the desired search targets.

[0041] According to one of the first to eighth methods of responding to retrieval requests through a KV storage device according to the third aspect of the present application, a ninth method of responding to retrieval requests through a KV storage device according to the third aspect of the present application is provided, and in response to the user's further selection of a search target, the original data storage location corresponding to the search target is obtained from the structured information, and the original data is obtained to be provided to the user.

[0042] According to one of the first to ninth methods of responding to retrieval requests through a KV storage device according to the third aspect of the present application, a tenth method of responding to retrieval requests through a KV storage device according to the third aspect of the present application is provided, and the original data is loaded in advance before the user makes further selection based on the original data storage location recorded in the structured information.

[0043] According to a fourth aspect of the present application, a program including program code is provided. When the program code is loaded into a storage device and executed on the storage device, the storage device executes one of the methods according to the first aspect or the third aspect of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0045] Figure 1 is a block diagram according to an embodiment of the present application;

[0046] Figure 2 is a schematic diagram of a machine learning component according to an embodiment of the present application;

[0047] Figure 3 It is a schematic diagram of responding to a retrieval request through a KV storage device according to an embodiment of the present application. DETAILED DESCRIPTION

[0048] The following is a clear and complete description of the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0049] Embodiment 1

[0050] Figure 1 It is a block diagram according to an embodiment of the present application.

[0051] Raw data is unstructured data such as videos, pictures, and text files.

[0052] The raw data is structured. For example, labels are added to the raw data to indicate the source, format, storage location, required access rights and other structured information of the raw data; the content of the raw data is analyzed, such as extracting thumbnails and summaries from videos or pictures, extracting objects such as portraits and cars, extracting facial features, height, gender, age and other features of people, extracting features such as car brands and license plates. The labels and features associated with the raw data are all structured information generated by the structured processing. Structured information is easy to retrieve, and structured information is retrieved through existing databases or search engine technologies, or other existing or future search technologies.

[0053] The structured information generated by the structured processing module 110 includes a variety of searchable elements (for example, objects contained in videos or pictures, features of objects, keywords in document summaries, etc.). Users can search these elements. For example, a user searches for pictures or videos containing people, and further specifies facial features, height, gender, and other features of the people as search targets. According to an embodiment of the present application, the search formula generation module 120 generates a variety of possible search formulas based on the searchable elements and combinations thereof provided by the structured processing module 110. The search formula includes one or more search elements to be searched.

[0054] Optionally, or further, the search formula generating module 120 further obtains the search behavior of the user, and obtains the search formula described by the user from the search behavior of the user.

[0055] The search formula generation module 120 uses the generated search formula to implement the search. For example, the search is implemented by providing the search formula to the search engine or database. The search engine or database 130 generates search results based on the search formula. Optionally, some search engines / databases 130 support search methods such as fuzzy search, semantic search, and image search. The search formula generation module 120 also uses these search methods to implement the search, and the search command provided to the search engine / database 130 during the search process is also called a search formula.

[0056] The search formula generation module 120 also provides the search formula for performing the search to the machine learning component 140 .

[0057] The machine learning component 140 is, for example, a machine learning component of the prior art or a machine learning component that will appear in the future.

[0058] The machine learning component 140 takes the search formula provided by the search formula generation module 120 as input. Optionally, each search element of the search formula is provided to each input node of the machine learning component 140 respectively. The machine learning component 140 generates a score for the search formula as output based on the input search formula. Optionally, the score is used as an evaluation of the quality of the search formula. For example, a relatively high score means that the search formula has a better search effect, while a relatively low score means that the search formula has a poor search effect and is not recommended for use. A better search effect, for example, can quickly and effectively obtain the search target from the search results, and there are fewer irrelevant contents in the search results; otherwise, it means a poor search effect.

[0059] Optionally, the machine learning component 140 is trained to generate similar scores for search terms with similar search results. For example, the search results of the implemented search are used as an evaluation of the score of the search terms output by the machine learning component 140, so that for search terms with similar search results, the machine learning component 140 outputs similar scores.

[0060] In the learning process of the machine learning component 140, the search formula is used as input, a score is generated based on the input, and the evaluation provided by the score evaluation module 150 is used as feedback on the generated score. The machine learning component 140 updates the weights of its internal nodes based on the evaluation to gradually generate a score that can more accurately evaluate the quality of the search formula.

[0061] The search formula input to the machine learning component 140 is also provided to the search engine / database 130. The search engine or database 130 generates search results based on the search formula. The scoring evaluation module 150 evaluates the scores generated by the machine learning component 140 based on the search results generated by the search engine or database 130, and provides the evaluation to the machine learning component 140. Optionally, the evaluation of the search formula is obtained through manual annotation, or the subsequent behavior of the user after performing the search using the search engine or database 130 is used as an evaluation of the score of the search formula output by the machine learning component 140, and the evaluation is provided to the machine learning component 140 (for example, if the user clicks on one or more search results, it means that the search formula has a good search effect, while if the user ignores all or most of the search results provided by the search engine or database 130, it means that the search formula has a poor search effect).

[0062] In another embodiment, the scoring and evaluation module 150 also classifies the search results of the search engine or database 130, and different classifications correspond to different values. The difference between the value corresponding to the category and the score of the machine learning component 140 is used as an evaluation of the score of the search formula output by the machine learning component 140, and the evaluation is provided to the machine learning component 140.

[0063] For the machine learning component 140 that has undergone the learning process, for the input search formula (denoted as s), the calculation results of the internal nodes of the machine learning component 140 are provided to the key generation (K generation) module 160. The K generation module 160 generates a key (K) for the KV storage device based on the calculation results of each node provided by the machine learning component 140. The key (K) corresponds to the search formula (s).

[0064] The result obtained by the search engine / database 130 processing the search formula (s) is provided to the value generation (V generation) module 170. The V generation module 170 generates a value (V) for the KV storage device based on the result provided by the search engine / database 130. The value (V) also corresponds to the search formula (s).

[0065] The key (K) and value (V) corresponding to the same search formula (s) are provided to the KV storage device, and the key (K) and value (V) are recorded in the KV storage device.

[0066] Figure 2It is a schematic diagram of a machine learning component according to an embodiment of the present application.

[0067] The machine learning component 140 includes, for example, an artificial neural network composed of multiple layers of nodes. Figure 2 The artificial neural network shown includes 4 layers (e.g., L0 layer, L1 layer, L2 layer, and L3 layer), L0 layer is the input layer, L3 layer is the output layer, L1 layer and L2 layer are internal layers. Each layer includes multiple nodes. For example, L0 layer includes nodes C1, C2...C5, L1 layer includes node 0, node 1, node 2, and node 3, L2 layer includes node 4, node 5, node 6, and node 7, and output layer L3 includes node 8 and node 9.

[0068] Each node of the input layer L0 receives an input search formula. For example, each node receives one of the search elements constituting the search formula.

[0069] The value of each node of the output layer L3 indicates the output of the machine learning component 140 (the score of the input search formula).

[0070] Each node in the internal layer represents the value calculated for the node coupled to it (denoted as N n , where n indicates the node in Figure 2 As an example:

[0071] N0=C1*F0(C1)+C2*F0(C2)+C3*F0(C3)+C4*F0(C4)+C5*F0(C5)

[0072] N1=C1*F1(C1)+C2*F1(C2)+C3*F1(C3)+C4*F1(C4)+C5*F1(C5), where

[0073] C1-C5 indicate inputs (quantized as values) received by each of the nodes C1-C5, F0() indicates a function associated with node 0, and F1() indicates a function associated with node 1. For other nodes in the artificial neural network, the values ​​corresponding to the nodes are obtained in a similar manner.

[0074] Thus, in response to a specified input (eg, search expression s), nodes 0 to 9 will generate respective values ​​(respectively denoted as N0 to N9).

[0075] Optionally, each node of the neural network in the prior art obtains the value of the node according to a variety of known functions.

[0076] Let (N0, N1, N2, N3) be the connection of the node values ​​N0, N1, N2 and N3. For example, if the node values ​​N0 to N3 are the numbers 0, 1, 2 and 3 respectively, then (N0, N1, N2, N3) is the digital sequence "0123". According to an embodiment of the present application, the node values ​​of each layer of the artificial neural network are connected respectively to obtain the sequences corresponding to the layers of the artificial neural network. For example, the sequence corresponding to the L1 layer is (N0, N1, N2, N3), the sequence corresponding to the L2 layer is (N4, N5, N6, N7), and the sequence corresponding to the L3 layer is (N8, N9).

[0077] Hash calculations are performed on the sequences corresponding to each layer to obtain hash values ​​corresponding to each layer. For example, the hash value corresponding to the L1 layer is Hash((N0, N1, N2, N3)), the hash value corresponding to the L2 layer is Hash((N4, N5, N6, N7)), and the hash value corresponding to the L3 layer is Hash((N8, N9)). Optionally, the hash values ​​corresponding to each layer have the same length.

[0078] The hash values ​​corresponding to one or more layers of the same artificial neural network are connected as the key (K) provided to the KV storage device. For example, the hash values ​​corresponding to the L2-L4 layers of the artificial neural network are connected to obtain (Hash((N0,N1,N2,N3)), Hash((N4,N5,N6,N7)), Hash((N8,N9))) as the key (K) provided to the KV storage device. Optionally, the connection result of the hash values ​​corresponding to one or more layers of the same artificial neural network (for example, (Hash((N0,N1,N2,N3)), Hash((N4,N5,N6,N7)), Hash((N8,N9)))) is hashed again, and the result is used as the key (K) provided to the KV storage device, so that the length of the key (K) is shortened or has a specified length.

[0079] See also Figure 1 , the K generation module 160 generates a key (K) provided to the KV storage device according to the value of one or more layers of nodes of the artificial neural network.

[0080] Return to view Figure 1 , the KV storage device records the pairs of keys (K) and values ​​(V) corresponding to the search formula (s). And the KV storage device responds to receiving a query request for the key (K) and outputs the value (V) corresponding to the key (K) as a response to the query request.

[0081] Embodiment 2

[0082] Figure 3 It is a schematic diagram of responding to a retrieval request through a KV storage device according to an embodiment of the present application.

[0083] The user provides a search formula for searching a database or a search engine, and the search formula includes one or more search elements. Optionally, the search formula is analyzed to extract the search elements from the search formula.

[0084] The search elements are provided to the machine learning component 140. See also e.g. Figure 2 , each search element is provided to each input node of the artificial neural network of the machine learning component 140. Each node of the artificial neural network of the machine learning component 140 generates its own node value in response to the input search element. The node value is provided to the K generation module 160. Figure 1 and Figure 2 The manner in which the K generation module 160 generates the key (K) is described, Figure 3 The K generation module 160 generates a key (K) according to the provided node value.

[0085] The generated key (K) is provided to the KV storage device for reading the value (V) corresponding to the key (K) from the KV storage device. The value (V) records the structured information corresponding to the key (K). The structured information also includes the storage location of the original data for generating the structured information. The structured information corresponding to the key (K) obtained from the KV storage device is used as a response to the search. Optionally, the structured information is displayed to the user as a result of the search. As an example, the structured information includes thumbnails of images or videos that are hit according to the features searched by the user. These thumbnails are displayed to the user so that the user can identify the desired search target. In response to the user's further selection of the search target, the original data storage location corresponding to the search target is obtained from the structured information, and the original data is obtained to be provided to the user.

[0086] Optionally, the original data is loaded in advance before the user makes further selections based on the original data storage location recorded in the structured information, thereby speeding up the response speed to the user's further selection of the search target.

[0087] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they are aware of the basic creative concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application. Obviously, those skilled in the art may make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for responding to a retrieval request through a KV storage device, comprising: Providing the search elements in the search formula to a machine learning component including an artificial neural network composed of multiple layers of nodes; wherein the artificial neural network includes an input layer, one or more internal layers, and an output layer, and each layer includes multiple nodes; The machine learning component provides the generated values ​​of one or more layers of nodes of its artificial neural network to the key generation component; The key generation component connects the nodes of one or more layers of the internal layer and output layer of the artificial neural network and outputs the corresponding hash value as the key, or performs hash calculation on the connection result of the hash value output of each node of one or more layers of the internal layer and output layer of the artificial neural network, uses the obtained result as the key, provides the generated key to the KV storage device, and in response to receiving a query request for the key, reads the value corresponding to the key from the KV storage device, and uses the value as a response to the query request.

2. The method according to claim 1, characterized in that The value corresponding to the same key records the structured information corresponding to the same key.

3. The method according to claim 2, characterized in that The structured information includes the storage location of the original data used to generate the structured information.

4. The method according to any one of claims 1 to 3, characterized in that The structured information corresponding to the key is obtained from the KV storage device as a response to the retrieval.

5. The method according to claim 4, characterized in that The method further comprises: Present structured information to users as a result of the search.

6. The method according to any one of claims 1 to 3 and 5, characterized in that The structured information includes thumbnails of images or videos that are hit according to the characteristics of the user's search.

7. The method according to claim 6, characterized in that Display thumbnails to users to help them identify the desired search target.

8. The method according to any one of claims 1 to 3, characterized in that In response to the user's further selection of a search target, the original data storage location corresponding to the search target is obtained from the structured information, and the original data is obtained to be provided to the user.

9. The method according to any one of claims 1 to 3, characterized in that The method further comprises: According to the original data storage location recorded in the structured information, the original data is loaded in advance before the user makes further selection.

10. A program comprising program code, which, when loaded into a storage device and executed on the storage device, causes the storage device to execute the method according to any one of claims 1 to 9.

Images