Information generating method, device, electronic device and computer-readable medium

Abstract

An information generating method, an apparatus, an electronic device, and a computer-readable medium are provided. The information generating method includes: obtaining a pre-constructed graph model related to a category of a target item, and serialized text data and structured text data related to the target item (401); performing word segmentation on the serialized text data to obtain a word set (402); matching target words in the word set with parameter names in the graph model, and generating a matched graph model based on the matching result (403); generating a fused graph model based on the structured text data and the matched graph model (404); and generating summary information of the target item based on the fused graph model (405). The method can accurately and effectively generate summary information of the target item by considering feature information of the serialized text data and the structured text data.

Description

[Technical field]

[0001] TECHNICAL FIELD The embodiments of the present disclosure relate to the field of computer technology, and in particular to information generating methods, apparatus, electronic devices, and computer-readable media. [Background technology]

[0002] At present, traditional technical applications of item search and recommendation cannot well meet the needs of an increasing number of users. When browsing a recommendation system, users often face the problem of information overload. Users want to quickly learn about products through some excellent item summaries. For generating item summaries, a commonly adopted method is to generate item summaries based on a recurrent neural network. However, a recurrent neural network can only model serial text information, and item information includes not only unstructured text sequences but also structured information. It is difficult for a recurrent neural network to directly model structured input, and current processing methods still process it as a string sequence, which may cause the loss of structured information and reduce the expressive ability of the model. Summary of the Invention [Problem to be solved by the invention]

[0003] The present disclosure is intended to introduce concepts in a simplified form, which are described in detail in the following specific embodiments. The present disclosure is not intended to identify key features or essential features of the claims, nor is it intended to limit the scope of the claims as claimed. [Means for solving the problem]

[0004] Some embodiments of the present disclosure provide an information generating method, an apparatus, a device, and a computer-readable medium to solve the technical problems described in the background art above.

[0005] As a first aspect, some embodiments of the present disclosure provide an information generating method, the method including: obtaining a pre-constructed graph model related to a category of a target item, serialized text data and structured text data related to the target item, the graph model representing an association relationship between a parameter name of the category of the target item and each parameter content information, a node in the graph model representing the parameter name or the parameter content information of the category of the target item, and a numerical value corresponding to an edge in the graph model representing association degree information between the parameter name and the parameter content information; performing word segmentation on the serialized text data to obtain a word set; matching target words in the word set with parameter names in the graph model, and generating a matching graph model based on the matching result; generating a fusion graph model based on the structured text data and the matching graph model; and generating summary information of the target item based on the fusion graph model.

[0006] Preferably, matching target words in the word set with parameter names in the graph model and generating a matched graph model based on the matching result includes: matching target words in the word set with parameter names in the graph model to obtain matching results; and generating a matched graph model based on the matching results and target edges in the graph model, where the target edges are edges representing relevance information between the target words and parameter names corresponding to the target words in the graph model.

[0007] Preferably, generating summary information of the target item based on the fused graph model includes converting the fused graph model into a corresponding vector matrix, identifying vectors corresponding to each node in the fused graph model, the vectors corresponding to each node representing feature information of the parameter name or the parameter content information, and generating summary information of the target item based on the vector matrix and the vectors corresponding to each node.

[0008] Preferably, generating summary information of the target item based on the vector matrix and the vector corresponding to each node includes: inputting the vector matrix and the vector corresponding to each node in the fused graph model into a pre-trained graph convolutional neural network to obtain a first output result; coding the first output result to obtain a second output result; and decoding the second output result to obtain the summary information of the target item.

[0009] Preferably, coding the first output result to obtain a second output result includes inputting the first output result into a coding network that is pre-trained to process time-series text data, and obtaining the second output result.

[0010] Preferably, decoding the second output result to obtain summary information corresponding to the target item includes inputting the second output result into a decoding network that is pre-trained to process time-series text data, to obtain summary information of the target item.

[0011] Preferably, the numerical values ​​corresponding to edges in the graph model are identified by the following steps: obtain item information of each item in the category of the target item; the item information includes a parameter name and at least one parameter numerical value; based on the item information of each item, identify each parameter content information corresponding to each parameter name and the number of times each parameter content information appears in the item information of each item; and based on the frequency of each parameter content information corresponding to each parameter name and the frequency of each parameter content information appearing, use a term frequency-inverse document frequency method to identify the numerical values ​​corresponding to edges in the graph model.

[0012] As a second aspect, some embodiments of the present disclosure provide an information generating device, the device including: an acquiring unit configured to acquire a pre-constructed graph model related to a category of a target item, serialized text data and structured text data related to the target item, where the graph model represents an association relationship between a parameter name of the category of the target item and each parameter numerical value, a node in the graph model represents a parameter name or parameter content information of the category of the target item, and a numerical value corresponding to an edge in the graph model represents an association degree information between the parameter name and the parameter content information; a word segmentation unit configured to perform word segmentation on the serialized text data to obtain a word set; a matching generation unit configured to match target words in the word set with parameter names in the graph model and generate a matched graph model based on a matching result; a first generation unit configured to generate a fused graph model based on the structured text data and the matched graph model; and a second generation unit configured to generate summary information of the target item based on the fused graph model.

[0013] Preferably, the matching generation unit further matches a target word in the word set with a parameter name in the graph model to obtain a matching result, and generates a graph model after matching based on the matching result and a target edge in the graph model, and the target edge is arranged as an edge in the graph model representing related degree information between the target word and a parameter name corresponding to the target word.

[0014] Preferably, the second generation unit is further configured to convert the fused graph model into a corresponding vector matrix, identify a vector corresponding to each node in the fused graph model, wherein the vector corresponding to each node represents feature information of the parameter name or the parameter content information, and generate summary information of the target item based on the vector matrix and the vector corresponding to each node.

[0015] Preferably, the second generation unit is further configured to input the vector matrix and the vector corresponding to each node in the fused graph model into a pre-trained graph convolutional neural network to obtain a first output result, code the first output result, obtain a second output result, decode the second output result, and obtain summary information of the target item.

[0016] Preferably, the second generation unit is further configured to input the first output result into a pre-trained coding network for processing time-series text data to obtain the second output result.

[0017] Preferably, the second generation unit is further configured to input the second output result into a pre-trained decoding network for processing time-series text data to obtain summary information of the target item.

[0018] As a third aspect, some embodiments of the present disclosure provide an electronic device, the electronic device including one or more processors and a storage device having one or more programs stored therein, the one or more programs, when executed by the one or more processors, causing the one or more processors to perform any of the methods of the first aspect.

[0019] In a fourth aspect, some embodiments of the present disclosure provide a computer-readable medium having a computer program stored thereon, the computer-readable medium implementing any of the methods of the first aspect when the program is executed by a processor. Effect of the Invention

[0020] The above embodiments of the present disclosure have the following beneficial effects, and the information generating method of some embodiments of the present disclosure can accurately and effectively generate summary information of a target item by considering the feature information of serialized text data and structured text data. Specifically, this is because the recurrent neural network can only model serial text information, and the item information includes not only unstructured text sequences but also structured information. It is difficult for the recurrent neural network to directly model structured input, and the current processing method still processes it as a string sequence, which may cause the loss of structured information and reduce the expressive ability of the model. Based on this, the information generating method of some embodiments of the present disclosure can first obtain a graph model related to the category of the target item, the serialized text data and the structured text data related to the target item, which are pre-constructed. In addition, the graph model represents the association relationship between the parameter name of the category of the target item and each parameter content information, the nodes in the graph model represent the parameter name or parameter content information of the category of the target item, and the numerical value corresponding to the edge in the graph model represents the association degree information between the parameter name and the parameter content information. Here, the graph model is used to identify parameter names corresponding to target words in a subsequent word set, which forms a basis for generating a subsequent matched graph model. The graph model is obtained to convert serialized text data related to the target item into structured data. Then, word segmentation is performed on the serialized text data to obtain a word set. Further, target words in the word set are matched with parameter names in the graph model, and a matched graph model is generated based on the matching result. Next, a fused graph model is generated based on the structured text data and the matched graph model.Wherein the fused graph model represents structured text information related to the target item. Finally, generate summary information of the target item based on the fused graph model. As can be seen, the information generating method can accurately and effectively generate summary information of the target item by considering the feature information of the serialized text data and the structured text data. [Brief description of the drawings]

[0021] The above and other features, advantages and aspects of each embodiment of the present disclosure will become more apparent with reference to the following specific embodiments with reference to the drawings. In the drawings, the same or similar reference numerals indicate the same or similar elements. The drawings are schematic, and the elements and components are not necessarily drawn to scale.

[0022] [Figure 1] FIG. 2 is a schematic diagram of an application scene of the information generation method of some embodiments of the present disclosure. [Diagram 2] FIG. 2 is a schematic diagram of an application scene of the information generation method of some embodiments of the present disclosure. [Diagram 3] FIG. 2 is a schematic diagram of an application scene of the information generation method of some embodiments of the present disclosure. [Figure 4] 1 is a flow chart of some exemplary information generation methods of the present disclosure. [Diagram 5] 5 is a flowchart of some further examples of information generation methods according to the present disclosure. [Figure 6] 1A to 1C are schematic diagrams illustrating configurations of some embodiments of an information generating device according to the present disclosure. [Figure 7] FIG. 1 is a schematic diagram of an electronic device suitable for implementing some embodiments of the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0023] The embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be realized in various forms and should not be construed as being limited to the embodiments described herein. On the contrary, these embodiments are provided to allow a more thorough and complete understanding of the present disclosure. It should be understood that the accompanying drawings and embodiments of the present disclosure are used only for illustrative purposes and are not used to limit the scope of protection of the present disclosure.

[0024] In addition, the drawings show only parts related to the present invention for the convenience of explanation. If not contradictory, the embodiments and features in the embodiments in the present disclosure can be combined with each other.

[0025] It should be noted that the concepts of "first", "second", etc. referred to in this disclosure are merely intended to distinguish different devices, modules, or units, and do not limit the order or interdependence of functions performed by these devices, modules, or units.

[0026] The modifications "one" and "multiple" referred to in this disclosure are exemplary rather than limiting, and should be understood by those of skill in the art as "one or more" unless the context clearly indicates otherwise.

[0027] The names of messages or information exchanged between multiple devices in the embodiments of the present disclosure are for illustrative purposes only and do not limit the scope of these messages or information.

[0028] Hereinafter, the present disclosure will be described in detail in accordance with embodiments with reference to the drawings.

[0029] 1 to 3 are schematic diagrams of application scenes of information generating methods according to some embodiments of the present disclosure.

[0030] 1 to 3, the electronic device 101 can first obtain a graph model 105 that is constructed in advance and is related to a category of a target item, serialized text data 104 that is related to a target item 102, and structured text data 103. The graph model 105 represents an association relationship between a parameter name of the category of the target item and each piece of parameter content information, a node in the graph model 105 represents a parameter name or parameter content information of the category of the target item, and a numerical value corresponding to an edge in the graph model 105 represents information on the degree of association between the parameter name and the parameter content information. In this application scenario, the graph model 105 includes: "Screen size->6.5 inches, relevance degree information: 0.6; Screen size->5.5 inches, relevance degree information: 0.4; Color->light red, relevance degree information: 0.4; Color->white, relevance degree information: 0.4; Color->black, relevance degree information: 0.3; Battery capacity->500, relevance degree information: 0.3; Battery capacity->mHh, relevance degree information: 0.9; Battery capacity->1000, relevance degree information: 0.6; Pixels->500, relevance degree information: 0.5; Pixels->1000, relevance degree information: 0.4; Pixels->800, relevance degree information: 0.5; Pixels->10,000, relevance degree information: 0.8". The target item 102 may be a "mobile phone". The serialized text data 104 may be "Mobile phone brand: **, color: red, battery capacity: 3000mAh, pixels: 10 million". The structured text data 103 may include "** mobile phone quality is good, screen size is large, light red, battery capacity is large, anti-slip and wear-resistant, high pixel". As shown in FIG. 2, the serialized text data 104 is subjected to word segmentation to obtain a word set 106. Furthermore, the target words 107 in the word set 106 are matched with the parameter names in the graph model 105, and a matching graph model 108 is generated based on the matching result.In this application scenario, the word set 106 may include “**mobile phone”, “quality”, “good”, “screen size”, “light red”, “large”, “capacity”, “large”, “slip-resistant”, “wear-resistant”, “pixels”, and “high”. The target word 107 may be “light red”. The matching graph model 108 may include “color->light red, relevance information: 0.4;**mobile phone; quality; good; screen size; capacity: large; wear-resistant: pixels; high”. As shown in FIG. 3, then, a fused graph model 109 is obtained based on the structured text data 103 and the matching graph model 108. Finally, summary information 110 of the target item 102 is generated based on the fused graph model 109. In this application scenario, the graph model 109 after fusion may include "color->light red, relevance information: 0.4; battery capacity->1000, relevance information: 0.6; pixels->1000, relevance information: 0.4; screen size->5.5 inches, relevance information: 0.4; ** mobile phone; quality; large; good; wear resistance; high". The summary information 110 may be "** mobile phone, good quality, large screen size, 5.5 inches, large battery capacity, 1000mAh, high pixels, 10 million".

[0031] The electronic device 101 may be hardware or software. If the electronic device is hardware, it may be implemented as a distributed cluster consisting of multiple servers or terminal devices, or as a single server or a single terminal device. If the electronic device is embodied as software, it may be installed in the hardware devices listed above. For example, it may be embodied as multiple software or software modules for providing distributed services, or as a single software or software module. No specific limitations are provided here.

[0032] It should be understood that the number of electronic devices shown in Figure 1 is merely exemplary, and any number of electronic devices may be included as desired.

[0033] Continuing to refer to Figure 4, there is shown a flow chart 400 of some embodiments of the information generation method of the present disclosure. The information generation method includes the following steps.

[0034] In step 401, a pre-constructed graph model related to a category of a target item, serialized text data related to the target item, and structured text data are obtained.

[0035] In some embodiments, an entity that executes the information generating method (for example, the electronic device 101 shown in FIG. 1) can acquire a graph model that is constructed in advance and is related to a category of a target item, serialized text data and structured text data related to the target item, by a wired or wireless connection method. The graph model represents an association relationship between a parameter name of the category of the target item and each parameter content information, a node in the graph model represents a parameter name or parameter content information of the category of the target item, and a numerical value corresponding to an edge in the graph model represents association degree information between the parameter name and the parameter content information. The association degree information may be a numerical value between 0 and 1.

[0036] In addition, the above wireless connection methods may include, but are not limited to, 3G / 4G / 5G connections, WiFi connections, Bluetooth (registered trademark) connections, WiMAX connections, Zigbee connections, UWB (ultra wideband) connections, and other currently known or future developed wireless connection methods.

[0037] In step 402, word segmentation is performed on the serialized text data to obtain a word set.

[0038] In some embodiments, the actor may perform word segmentation on the serialized text data to obtain a set of words. Illustratively, the actor may perform word segmentation on the serialized text data to obtain a set of words using stuttering word segmentation.

[0039] In step 403, the target words in the word set are matched with the parameter names in the graph model, and a matched graph model is generated based on the matching result.

[0040] In some embodiments, the execution entity can match target words in the word set with parameter names in the graph model, and generate a matched graph model based on the matching result. The target words can also be words that are related to the parameter names in the graph model. For example, the execution entity can convert the matching result into a graph format, and obtain a matched graph model.

[0041] In some alternative implementation forms of some embodiments, matching target words in the word set with parameter names in the graph model, and generating a matched graph model based on the matching results, may include the following steps:

[0042] In the first step, a target word in the word set is matched with a parameter name in the graph model to obtain a matching result. For example, the execution entity can first extract semantic information of the target word in the word set, then identify a parameter name related to the semantic information of the target word by traversing each parameter name in the graph model, and finally match a parameter name related to the semantic information of the target word with the target word to obtain a matching result.

[0043] In a second step, a graph model after matching is generated based on the matching result and a target edge in the graph model. The target edge is an edge that represents information on the degree of association between the target word and a parameter name corresponding to the target word in the graph model. For example, the execution subject can specify node information of the graph model after matching based on the matching result. Finally, a graph model after matching is generated by setting the target edge in the graph model as an edge of the graph model after matching.

[0044] Step 404: generating a fused graph model based on the structured text data and the matched graph model.

[0045] In some alternative implementations of some embodiments, the execution entity can generate a fused graph model based on the structured text data and the matched graph model. For example, the execution entity can fuse the structured text data with the matched graph model to obtain a fused graph model, where the fused graph model represents structured text information related to a target item.

[0046] Step 405: Generate summary information of the target item based on the fused graph model.

[0047] In some embodiments, the execution entity can generate summary information of the target item based on the fused graph model. For example, the execution entity can first convert the fused graph model into corresponding vectors, and then input the corresponding vectors and vectors corresponding to each node in the fused graph model into a pre-trained recurrent neural network to obtain summary information of the target item.

[0048] In some alternative implementations of some embodiments, the values ​​corresponding to edges in the graph model are identified by the following steps.

[0049] A first step is to obtain item information for each item in the category of the target item, and the item information includes a parameter name and at least one parameter value.

[0050] In a second step, each parameter content information corresponding to each parameter name and the number of times each parameter content information appears in the item information of each item are identified based on the item information of each item. Here, each item information in the item information of each item corresponds to at least one parameter name. Furthermore, a parameter name set corresponding to the item information of each item can be identified. Also, there is a possibility that a plurality of duplicated parameter names exist in the parameter name set. As a result, duplication removal can be performed on the parameter name set to obtain a parameter name set after duplication removal. Here, each parameter name in the parameter name set after duplication removal corresponds to at least one parameter content information. As a result, the number of times each parameter content information appears in the item information of each item can be identified by the item information of each item.

[0051] In the third step, a numerical value corresponding to an edge in the graph model is identified using a term frequency-inverse document frequency method based on each parameter content information corresponding to each parameter name and the frequency at which each parameter content information appears.

[0052] Here, the main idea of ​​the term frequency-inverse document frequency method is that if a word or phrase has a high frequency (Term Frequency, TF) in one text and does not appear often in other texts, the word or phrase is considered to have good category classification ability and is suitable for classification. Inverse document frequency (IDF) mainly means that the fewer documents that contain a word or phrase, the larger the IDF will be, and the better the category classification ability of the word or phrase. Thus, the importance of a word or phrase in a text can be calculated by using the term frequency-inverse document frequency method.

[0053] The above embodiments of the present disclosure have the following beneficial effects, and the information generating method of some embodiments of the present disclosure can accurately and effectively generate summary information of a target item by considering the feature information of serialized text data and structured text data. Specifically, this is because the recurrent neural network can only model serial text information, and the item information includes not only unstructured text sequences but also structured information. It is difficult for the recurrent neural network to directly model structured input, and the current processing method still processes it as a string sequence, which may cause the loss of structured information and reduce the expressive ability of the model. Based on this, the information generating method of some embodiments of the present disclosure can first obtain a graph model related to the category of the target item, the serialized text data and the structured text data related to the target item, which are pre-constructed. In addition, the graph model represents the association relationship between the parameter name of the category of the target item and each parameter content information, the nodes in the graph model represent the parameter name or parameter content information of the category of the target item, and the numerical value corresponding to the edge in the graph model represents the association degree information between the parameter name and the parameter content information. Here, the graph model is used to identify parameter names corresponding to target words in a subsequent word set, which forms a basis for generating a subsequent matched graph model. The graph model is obtained to convert serialized text data related to the target item into structured data. Then, word segmentation is performed on the serialized text data to obtain a word set. Further, target words in the word set are matched with parameter names in the graph model, and a matched graph model is generated based on the matching result. Next, a fused graph model is generated based on the structured text data and the matched graph model.Wherein the fused graph model represents structured text information related to the target item. Finally, generate summary information of the target item based on the fused graph model. As can be seen, the information generating method can accurately and effectively generate summary information of the target item by considering the feature information of the serialized text data and the structured text data.

[0054] 5, there is shown a flow chart 500 of some other embodiments of the information generation method of the present disclosure. The information generation method includes the following steps.

[0055] In step 501, a pre-constructed graph model related to a category of a target item, serialized text data related to the target item, and structured text data are obtained.

[0056] In step 502, word segmentation is performed on the serialized text data to obtain a word set.

[0057] In step 503, the target words in the word set are matched with the parameter names in the graph model, and a matched graph model is generated based on the matching result.

[0058] In step 504, a fused graph model is generated based on the structured text data and the matched graph model.

[0059] In some embodiments, the specific implementation and technical effects of steps 501-504 may be referred to steps 401-404 in those embodiments corresponding to FIG. 4, and will not be repeated here.

[0060] In step 505, the fused graph model is converted into a corresponding vector matrix.

[0061] In some embodiments, an execution entity (e.g., the electronic device 101 shown in FIG. 1) can convert the fused graph model into a corresponding vector matrix. For example, the execution entity can generate a corresponding vector matrix based on a numerical value corresponding to each edge in the fused graph model. In addition, the relationship between an isolated node and other nodes can be represented by a numerical value of 0.

[0062] Step 506 identifies vectors corresponding to each node in the fused graph model.

[0063] In some embodiments, the execution entity can identify vectors corresponding to each node in the fused graph model, and the vectors corresponding to each node represent feature information of the parameter names or the parameter content information. For example, the parameter names or parameter content information corresponding to each node in the fused graph model can be converted into corresponding vectors through a word embedding method.

[0064] In step 507, the vector matrix and the vectors corresponding to each node in the fused graph model are input into a pre-trained graph convolutional neural network to obtain a first output result.

[0065] In some embodiments, the execution entity may input the vector matrix and the vectors corresponding to each node in the graph model after the fusion into a pre-trained graph convolutional neural network (GCN) to obtain a first output result. The graph convolutional network may be a multi-layer graph convolutional layer. The graph convolutional layer may aggregate the neighbor information of each node through an adjacency matrix, further enhancing the characterization of the current node, and improving the modeling accuracy for the model input. By performing multi-layer graph convolutional layer modeling, multi-hop neighbor information may be obtained.

[0066] Step 508: coding the first output result to obtain a second output result.

[0067] In some embodiments, the execution entity can code the first output result to obtain a second output result. For example, the execution entity can input the first output result into a pre-trained coding network to obtain a second output result.

[0068] In some alternative implementations of some embodiments, the execution entity may input the first output result into a coding network that is pre-trained and processes time-series text data to obtain the second output result, and the coding network that processes the time-series text data may be one of a Recurrent Neural Network (RNN) and a Long Short-Term Memory (LSTM) network.

[0069] Step 509: Decoding the second output result to obtain summary information of the target item.

[0070] In some embodiments, the execution entity can decode the second output result to obtain summary information of the target item. For example, the execution entity can input the second output result into a pre-trained decoding network to obtain the second output result.

[0071] In some alternative implementations of some embodiments, the execution entity inputs the second output result into a pre-trained decoding network for processing time-series text data to obtain summary information of the target item, and the decoding network for processing time-series text data may be one of a recurrent neural network and a long short-term memory network.

[0072] As can be seen from Fig. 5, compared with the description of some embodiments corresponding to Fig. 4, the flow 500 of the information generating method in some embodiments corresponding to Fig. 5 illustrates a step of generating summary information of the target items based on the fused graph model, so that the technical solutions described in these embodiments can generate summary information of the target items more accurately and effectively.

[0073] Continuing to refer to FIG. 6, as an implementation of the method described in each of the above figures, the present disclosure provides several embodiments of information generating devices, which correspond to the embodiments of the method described in FIG. 4, and which may be specifically applied to various electronic devices.

[0074] As shown in Fig. 6, an information generating device 600 of some embodiments includes an acquiring unit 601, a word segmentation unit 602, a matching generation unit 603, a first generating unit 604 and a second generating unit 605. The acquiring unit 601 is configured to acquire a pre-constructed graph model related to a category of a target item, serialized text data related to the target item and structured text data, the graph model represents an association relationship between a parameter name of the category of the target item and each parameter value, a node in the graph model represents a parameter name or parameter content information of the category of the target item, and a value corresponding to an edge in the graph model represents association degree information between the parameter name and the parameter content information. The word segmentation unit 602 is configured to perform word segmentation on the serialized text data to obtain a word set. The matching generation unit 603 is configured to match a target word in the word set with a parameter name in the graph model, and generate a matching graph model according to a matching result. The first generating unit 604 is configured to generate a fused graph model based on the structured text data and the matching graph model, and the second generating unit 605 is configured to generate summary information of the target item based on the fused graph model.

[0075] In some optional implementation forms of some embodiments, the matching generation unit 603 of the information generating device 600 may be further configured to match a target word in the word set with a parameter name in the graph model to obtain a matching result, and generate a post-matching graph model based on the matching result and a target edge in the graph model, and the target edge is an edge representing related degree information between the target word and a parameter name corresponding to the target word in the graph model.

[0076] In some optional implementation forms of some embodiments, the second generation unit 605 of the information generation device 600 may be further configured to convert the fused graph model into a corresponding vector matrix, identify vectors corresponding to each node in the fused graph model, and the vectors corresponding to each node represent feature information of the parameter name or the parameter content information, and generate summary information of the target item based on the vector matrix and the vectors corresponding to each node.

[0077] In some optional implementation forms of some embodiments, the second generation unit 605 of the information generating device 600 may be further configured to input the vector matrix and the vector corresponding to each node in the fused graph model into a pre-trained graph convolutional neural network to obtain a first output result, code the first output result, obtain a second output result, decode the second output result, and obtain summary information of the target item.

[0078] In some alternative implementation forms of some embodiments, the second generation unit 605 of the information generating device 600 may be further configured to input the first output result into a pre-trained coding network for processing time-series text data to obtain the second output result.

[0079] In some alternative implementation forms of some embodiments, the second generation unit 605 of the information generating device 600 may be further configured to input the second output result into a pre-trained decoding network for processing time-series text data to obtain summary information of the target item.

[0080] As can be seen, the units described in the apparatus 600 correspond to the steps in the method described with reference to Fig. 4. Thus, the operations, features and advantageous effects achieved as described for the above-mentioned method apply to the apparatus 600 and the units contained therein as well and will not be repeated here.

[0081] 7, which shows a schematic diagram of an electronic device 700 for implementing some embodiments of the present disclosure. The electronic device shown in FIG. 7 is merely an example and should not impose any limitations on the functions and scope of use of the embodiments of the present disclosure.

[0082] 7, the electronic device 700 may include a processing unit (e.g., a central processing unit, a graphics processor, etc.) 701, which may perform various suitable operations and processes according to a program stored in a read only memory (ROM) 702 or a program loaded from a storage device 707 into a random access memory (RAM) 703. The RAM 703 further stores various programs and data necessary for the operation of the electronic device 700. The processing unit 701, the ROM 702, and the RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0083] Typically, the following devices may be connected to the I / O interface 705: input devices 706 including, for example, a touch screen, touch pad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; output devices 707 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 708 including, for example, a tape, hard disk, etc.; and communication devices 709. The communication devices 709 may allow the electronic device 700 to communicate wirelessly or wired with other devices to exchange data. Although the electronic device 700 is shown in FIG. 7 with various devices, it is understood that it is not required to implement or include all the devices shown. Alternatively, more or fewer devices may be implemented or included. Each block shown in FIG. 7 may represent one device, but may also represent multiple devices if desired.

[0084] In particular, according to some embodiments of the present disclosure, the processes described with reference to the above flowcharts can be implemented as computer software programs. For example, some embodiments of the present disclosure include a computer program product, which includes a computer program carried on a computer-readable medium, the computer program including program code for performing the methods shown in the flowcharts. In some such embodiments, the computer program can be downloaded and installed from a network via the communication device 709, or installed from the storage device 708, or installed from the ROM 702. When the computer program is executed by the processing device 701, it performs the functions limited to the methods of some embodiments of the present disclosure.

[0085] It should be noted that the computer-readable medium in some embodiments of the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium may be, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any combination of the above. In some embodiments of the present disclosure, the computer-readable storage medium may be a tangible medium that includes or stores a program, which can be instructed to perform the use of or in conjunction with a system, apparatus, or device. In some embodiments of the present disclosure, the computer-readable signal medium may include a data signal propagating in baseband or as part of a carrier wave, with computer-readable program code embedded therein. Such propagated data signals may take a number of forms, including, but not limited to, electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may be transmitted, propagated, or transmitted with instructions to execute a program for use in or in connection with a system, device, or appliance. The program code contained in the computer-readable medium may be transmitted over any suitable medium, including, but not limited to, wire, fiber optics, RF (radio frequency), or the like, or any suitable combination of the above.

[0086] In some embodiments, clients and servers may communicate using any now known or later developed network protocol, such as HyperText Transfer Protocol (HTTP), and may interconnect with any form or medium of digital data communication (e.g., a communications network). Examples of communications networks include local area networks ("LANs"), wide area networks ("WANs"), extranets (e.g., the Internet), and end-to-end networks (e.g., ad-hoc end-to-end networks), and any now known or later developed networks.

[0087] The computer-readable medium may be included in the device, or may exist independently and not be implemented in the electronic device. The computer-readable medium is loaded with one or more programs, and when the one or more programs are executed by the electronic device, the electronic device is caused to execute the following: obtain a graph model related to a category of a target item, serialized text data and structured text data related to the target item, the graph model represents an association relationship between a parameter name of the category of the target item and each parameter content information, a node in the graph model represents a parameter name or parameter content information of the category of the target item, and a value corresponding to an edge in the graph model represents an association degree information between the parameter name and the parameter content information, perform word segmentation on the serialized text data to obtain a word set, match a target word in the word set with a parameter name in the graph model, and generate a matching graph model based on a matching result, generate a fusion graph model based on the structured text data and the matching graph model, and generate summary information of the target item based on the fusion graph model.

[0088] Computer program code for carrying out the operations of some embodiments of the present disclosure can be programmed in one or more program design languages ​​or combinations thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and even conventional procedural programming languages ​​such as "C" or similar. The program code can run entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer by any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., providing an Internet connection through an Internet service provider).

[0089] The flowcharts and block diagrams in the accompanying drawings illustrate possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowcharts or block diagrams may represent a module, program segment, or part of code, which includes executable instructions for implementing one or more predetermined logical functions. It should also be noted that in some alternative implementations, the functions depicted in the blocks may occur in a different order than the order depicted in the drawings. For example, two blocks shown in succession may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functionality involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented in a system based on dedicated hardware that performs a predetermined function or operation, or may be implemented in a combination of dedicated hardware and computer instructions.

[0090] The units described in some embodiments of the present disclosure may be implemented in the form of software or hardware. The described units may be installed in a processor, and may be described, for example, as follows, in which the processor includes an acquisition unit, a word segmentation unit, a matching generation unit, a first generation unit, and a second generation unit. In addition, the names of these units may not constitute limitations on the units themselves in some circumstances, and for example, the word segmentation unit may be further described as "a unit for performing word segmentation on the serialized text data to obtain a word set."

[0091] The functions described herein above may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), etc.

[0092] The above is a description of some preferred embodiments of the present disclosure and the technical principles that apply. Those skilled in the art should understand that the scope of the invention according to the embodiments of the present disclosure is not limited to the technical solutions consisting of the specific combinations of the above technical features, but should also include other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the concept of the present invention. For example, the technical solutions formed by interchanging the above features with technical features having similar functions (but not limited to) disclosed in the embodiments of the present disclosure.

Claims

1. A method of generating information for use in an electronic device having a processor, comprising: The processor, Obtaining a graph model, which is constructed in advance and is related to a category of a target item, serialized text data and structured text data related to the target item, the graph model representing an association relationship between a parameter name of the category of the target item and each parameter content information, a node in the graph model representing a parameter name or parameter content information of the category of the target item, and a numerical value corresponding to an edge in the graph model representing association degree information between the parameter name and the parameter content information; performing word segmentation on the serialized text data to obtain a set of words; Matching target words in the word set with parameter names in the graph model, and converting the matching result into a graph format, thereby obtaining a matched graph model; fusing the structured text data with the matched graph model to obtain a fused graph model, where the fused graph model represents structured text information related to the target item; converting the fused graph model into a corresponding vector matrix; identifying a vector corresponding to each node in the fused graph model, where the vector corresponding to each node represents feature information of the parameter name or the parameter content information; and generating summary information of the target item based on the vector matrix and the vector corresponding to each node.

2. Matching target words in the word set with parameter names in the graph model, and generating a matched graph model based on a matching result, matching target words in the word set with parameter names in the graph model to obtain matching results; The method according to claim 1, further comprising: generating a graph model after matching based on the matching result and a target edge in the graph model, the target edge being an edge representing relevance information between the target word and a parameter name corresponding to the target word in the graph model.

3. generating summary information of the target item based on the vector matrix and the vector corresponding to each node, inputting the vector matrix and vectors corresponding to each node in the fused graph model into a pre-trained graph convolutional neural network to obtain a first output result; coding the first output result to obtain a second output result; 2. The method of claim 1, further comprising: decoding the second output result to obtain summary information of the target item.

4. Coding the first output result to obtain a second output result includes:

4. The method of claim 3, further comprising inputting the first output result into a coding network that is pre-trained to process time-series text data to obtain the second output result.

5. Decoding the second output result to obtain summary information corresponding to the target item includes: The method according to claim 3 or 4, further comprising inputting the second output result into a decoding network that is pre-trained to process time-series text data, to obtain summary information of the target item.

6. Numeric values ​​corresponding to edges in the graph model are identified by the following steps: The method according to any one of claims 1 to 5, further comprising the steps of: obtaining item information for each item in the category of the target item; the item information including a parameter name and at least one parameter numerical value; determining, based on the item information for each item, parameter content information corresponding to each parameter name and the number of times each parameter content information appears in the item information for each item; and determining, based on the frequency at which each parameter content information corresponding to each parameter name and the frequency at which each parameter content information appears, a numerical value corresponding to an edge in the graph model using a term frequency-inverse document frequency method.

7. An information generating device, an acquiring unit configured to acquire a graph model related to a category of a target item, serialized text data and structured text data related to the target item, the graph model representing an association relationship between a parameter name of the category of the target item and each parameter value, a node in the graph model representing a parameter name or parameter content information of the category of the target item, and a value corresponding to an edge in the graph model representing association degree information between the parameter name and the parameter content information; a word segmentation unit arranged to perform word segmentation on the serialized text data to obtain a set of words; a matching generation unit for matching target words in the word set with parameter names in the graph model, and converting the matching result into a graph format, thereby obtaining a matching graph model; a first generation unit for fusing the structured text data with the matched graph model to obtain a fused graph model, where the fused graph model is configured to represent structured text information related to the target item; a second generation unit configured to convert the fused graph model into a corresponding vector matrix, identify a vector corresponding to each node in the fused graph model, where the vector corresponding to each node represents feature information of the parameter name or the parameter content information, and generate summary information of the target item based on the vector matrix and the vector corresponding to each node.

8. An electronic device, An electronic device comprising one or more processors and a storage device for storing one or more programs, the electronic device causing the one or more processors to perform the method according to any one of claims 1 to 6 when the one or more programs are executed by the one or more processors.

9. 1. A computer-readable medium, comprising: A computer program is stored on the computer readable medium, and when the program is executed by a processor, the method according to any one of claims 1 to 6 is realized.

Images