Information generation method and apparatus, electronic device, and computer readable medium

Abstract

Embodiments of the present disclosure disclose information generation methods, devices, electronic equipment and computer readable media. A specific embodiment of the information generation method comprises: obtaining a pre-constructed graph model associated with a target item category, serialized text data related to a target item, and structured text data; performing word segmentation on the serialized text data to obtain a word set; matching a target word in the word set with a parameter name in the graph model and generating a matched graph model according to a matching result; generating a fused graph model according to the structured text data and the matched graph model; and generating summary information of the target item according to the fused graph model. The embodiment can accurately and effectively generate summary information of the target item by considering the characteristic information of the serialized text data and the structured text data.

Description

TECHNICAL FIELD

[0001] Embodiments of the present disclosure relate to the technical field of computer, and in particular, to an information generation method and device, an electronic device and a computer readable medium. BACKGROUND

[0002] At present, the traditional item search and recommendation technology application cannot well meet the growing needs of users. When browsing the recommendation system, users often face the problem of information explosion. Users want to quickly understand the product through some good item abstracts. For the generation of item abstracts, the commonly used way is to generate the abstract of the item based on a recurrent neural network. However, the recurrent neural network can only model sequence text information. Since the item information not only includes unstructured text sequence, but also includes structured information. The recurrent neural network is difficult to directly model the structured input. The current processing method is still to process it as a string sequence. This processing method may lose structured information, thereby reducing the expression ability of the model. SUMMARY

[0003] This part of the disclosure is used to introduce the concept in a brief form, which will be described in detail in the specific implementation part. This part of the disclosure is not intended to identify the key features or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0004] Some embodiments of the present disclosure provide an information generation method, device, equipment and computer readable medium to solve the technical problems mentioned in the background part.

[0005] In a first aspect, some embodiments of the present disclosure provide an information generation method, which comprises: obtaining a pre-constructed graph model associated with a target item category, sequence text data related to a target item and structured text data, wherein the graph model represents the association relationship between the parameter name and the parameter content information of the target item category, the nodes in the graph model represent the parameter name or the parameter content information of the target item category, 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; performing word segmentation on the sequence text data to obtain a word set; matching the target word in the word set with the parameter name in the graph model and generating a matched graph model according to the matching result; generating a fused graph model according to the structured text data and the matched graph model; and generating abstract information of the target item according to the fused graph model.

[0006] Optionally, the matching the target word in the word set with the parameter name in the graph model and generating a matched graph model according to the matching result comprises: matching the target word in the word set with the parameter name in the graph model to obtain a matching result; and generating a matched graph model according to the matching result and a target edge in the graph model, wherein the target edge is an edge in the graph model representing the association degree information between the target word and the parameter name corresponding to the target word.

[0007] Optionally, the generating the summary information of the target item according to the fused graph model comprises: converting the fused graph model into a corresponding vector matrix; determining 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 generating the summary information of the target item according to the vector matrix and the vector corresponding to each node.

[0008] Optionally, the generating the summary information of the target item according to the vector matrix and the vector corresponding to each node comprises: 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; encoding 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] Optionally, the encoding the first output result to obtain the second output result comprises: inputting the first output result into a pre-trained encoding network for processing time-series text data to obtain the second output result.

[0010] Optionally, the decoding the second output result to obtain the summary information of the target item comprises: inputting the second output result into a pre-trained decoding network for processing time-series text data to obtain the summary information of the target item.

[0011] Optionally, the value corresponding to the edge in the graph model is determined by: obtaining item information of each item in the category of the target item, wherein the item information comprises a parameter name and at least one parameter content information; determining, according to the item information of each item, each parameter content information corresponding to each parameter name and the number of occurrences of each parameter content information in the item information of each item; and determining, according to each parameter content information corresponding to each parameter name and the frequency of occurrence of each parameter content information, the value corresponding to the edge in the graph model by using a term frequency-inverse document frequency method.

[0012] In a second aspect, some embodiments of the present disclosure provide an information generation apparatus. The apparatus comprises: an acquisition unit configured to acquire a pre-constructed graph model associated with a target item category, serialized text data related to a target item, and structured text data, wherein the graph model represents an association relationship between a parameter name and each parameter content information of the target item category, a node in the graph model represents a parameter name or a parameter content information of the target item category, and a value corresponding to an edge in the graph model represents 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 a target word in the word set with a parameter name in the graph model and generate a matched graph model according to a matching result; a first generation unit configured to generate a fused graph model according to the structured text data and the matched graph model; and a second generation unit configured to generate summary information of the target item according to the fused graph model.

[0013] Optionally, the matching generation unit is further configured to: match the target word in the word set with the parameter name in the graph model to obtain a matching result; and generate the matched graph model according to the matching result and a target edge in the graph model, wherein the target edge is an edge in the graph model representing the association degree information between the target word and the parameter name corresponding to the target word.

[0014] Optionally, the second generation unit is further configured to: convert the fused graph model into a corresponding vector matrix; determine 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 the summary information of the target item according to the vector matrix and the vector corresponding to each node.

[0015] Optionally, 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; encode the first output result to obtain a second output result; and decode the second output result to obtain the summary information of the target item.

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

[0017] Optionally, 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 the summary information of the target item.

[0018] In a third aspect, some embodiments of the present disclosure provide an electronic device, comprising: one or more processors; a storage device having one or more programs stored thereon, when the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one 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, wherein the program, when executed by a processor, implements the method of any one of the first aspect.

[0020] The above-mentioned various embodiments of the present disclosure have the following beneficial effects: The information generation method of some embodiments of the present disclosure can accurately and effectively generate the summary information of the target item by considering the feature information of the serialized text data and the structured text data. Specifically, the recurrent neural network can only model the sequence text information, and since the item information includes not only unstructured text sequences but also structured information. The recurrent neural network is difficult to model directly for structured input, and the current processing method is still to process as a string sequence. This processing method may lose structured information, thereby reducing the expression ability of the model. Based on this, the information generation method of some embodiments of the present disclosure can first obtain a pre-constructed graph model associated with the target item category, serialized text data related to the target item, and structured text data. Wherein, the graph model represents the association relationship between the parameter name of the target item category and each parameter content information, the nodes in the graph model represent the parameter name or parameter content information of the target item category, 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 determine the parameter name corresponding to the target word in the subsequent word set, and lays the foundation for subsequent generation of the matched graph model. It should be noted that the graph model is used to convert the serialized text data related to the target item into structured data. Then, the serialized text data is segmented to obtain a word set. Further, the target word in the word set and the parameter name in the graph model are matched, and a matched graph model is generated according to the matching result. Then, according to the structured text data and the matched graph model, a fused graph model is generated. Here, the fused graph model embodies the structured text information related to the target item. Finally, according to the fused graph model, the summary information of the target item is generated. Thus, the information generation method can accurately and effectively generate the summary information of the target item by considering the feature information of the serialized text data and the structured text data. BRIEF DESCRIPTION OF DRAWINGS

[0021] The above and other features, advantages, and aspects of embodiments of the present disclosure will become more apparent by describing in detail some embodiments thereof with reference to the annexed drawings. In the drawings: like reference numerals refer to like elements throughout. It should be noted that the drawings are schematic and elements in the drawings are not necessarily to scale.

[0022] Figures 1-3 is a schematic diagram of one application scenario of the information generation method of some embodiments of the present disclosure;

[0023] Figure 4 is a flowchart of some embodiments of the information generation method according to the present disclosure;

[0024] Figure 5 is a flowchart of other embodiments of the information generation method according to the present disclosure;

[0025] Figure 6 is a structural schematic diagram of some embodiments of the information generation apparatus according to the present disclosure;

[0026] Figure 7 is a structural schematic diagram of an electronic device suitable for use to implement some embodiments of the present disclosure. DETAILED DESCRIPTION

[0027] Embodiments of the present disclosure will be described in more detail with reference to the drawings. Although some embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure can be more thoroughly and completely understood. It should be understood that the drawings and embodiments of the present disclosure are only for illustrative purposes and are not intended to limit the scope of protection of the present disclosure.

[0028] It should also be noted that, for ease of description, only parts related to the present disclosure are shown in the drawings. The embodiments in the present disclosure and the features in the embodiments can be combined with each other without conflict.

[0029] It should be noted that the terms “first”, “second”, and the like mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not intended to limit the order or interdependence of the functions performed by these devices, modules or units.

[0030] It should be noted that the terms “one”, “multiple” mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise explicitly stated in the context, it should be understood as “one or more”.

[0031] Names of messages or information exchanged between multiple devices in the embodiments of the present disclosure are only for illustrative purposes, and are not intended to limit the scope of the messages or information.

[0032] The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0033] Figures 1-3 is a schematic diagram of one application scenario of an information generation method of some embodiments of the present disclosure.

[0034] As Figures 1-3 shown, the electronic device 101 can first obtain a pre-constructed graph model 105 associated with a target item category, serialized text data 104 and structured text data 103 related to a target item 102. Among them, the above-mentioned graph model 105 represents the association relationship between the parameter name and the parameter content information of the above-mentioned target item category, the node in the above-mentioned graph model 105 represents the parameter name or parameter content information of the above-mentioned target item category, and the value corresponding to the edge in the above-mentioned graph model 105 represents the association degree information between the above-mentioned parameter name and the above-mentioned parameter content information. In this application scenario, the above-mentioned graph model 105 includes: "screen size -> 6.5 inches, association degree information: 0.6; screen size -> 5.5 inches, association degree information: 0.4; color -> light red, association degree information: 0.4; color -> white, association degree information: 0.4; color -> black, association degree information: 0.3, battery capacity -> 500, association degree information: 0.3; battery capacity -> mHh, association degree information: 0.9; battery capacity -> 1000, association degree information: 0.6; pixel -> 500, association degree information: 0.5; pixel -> 1000, association degree information: 0.4; pixel -> 800, association degree information: 0.5; pixel -> million, association degree information: 0.8". The above-mentioned target item 102 can be: "mobile phone". The above-mentioned serialized text data 104 can be: "mobile phone brand: **, color: red, battery capacity: 3000mAh, pixel: 1000 million". The structured text data 103 can include: "** mobile phone quality is good, screen size is large, light red, battery capacity is large, anti-slip and wear-resistant, high pixel". As Figure 2As shown, the serialized text data 104 is then segmented to obtain a word set 106. Next, the target word 107 in the word set 106 is matched with the parameter names in the graph model 105, and a matched graph model 108 is generated based on the matching results. In this application scenario, the word set 106 may include: "**phone", "quality", "good", "screen size", "light red", "large", "capacity", "large", "anti-slip", "wear-resistant", "pixels", "high". The target word 107 may be: "light red". The matched graph model 108 may include: "color->light red, association information: 0.4; **phone; quality; good; screen size; capacity: large; wear-resistant: pixels; high". Figure 3 As shown, next, based on the structured text data 103 and the matched graph model 108, a fused graph model 109 is obtained. Finally, based on the fused graph model 109, a summary information 110 of the target item 102 is generated. In this application scenario, the fused graph model 109 may include: "Color -> Light red, correlation information: 0.4; Battery capacity -> 1000, correlation information: 0.6; Pixels -> 1000, correlation information: 0.4; Screen size -> 5.5 inches, correlation information: 0.4; **Mobile phone; Quality; Large; Good; Wear-resistant; High". The summary information 110 may be: "**Mobile phone, good quality, large screen size, 5.5 inches, large battery capacity, 1000mAh, high pixel count, 10 million."

[0035] It should be noted that the aforementioned electronic device 101 can be either hardware or software. When the electronic device is hardware, it can be implemented as a distributed cluster consisting of multiple servers or terminal devices, or as a single server or a single terminal device. When the electronic device is software, it can be installed in the hardware devices listed above. It can be implemented as, for example, multiple software programs or software modules used to provide distributed services, or as a single software program or software module. No specific limitations are made here.

[0036] It should be understood that Figure 1 The number of electronic devices shown is merely illustrative. Any number of electronic devices can be used depending on the implementation requirements.

[0037] Continue to refer to Figure 4 The flowchart 400 illustrates some embodiments of the information generation method according to this disclosure. The information generation method includes the following steps:

[0038] Step 401: Obtain the pre-built graph model associated with the target item category, the serialized text data related to the target item, and the structured text data.

[0039] In some embodiments, the execution subject of the information generation method (e.g. Figure 1 The electronic device 101 shown in the figure) can obtain a pre-constructed graph model associated with the target item category, serialized text data related to the target item, and structured text data through wired connection or wireless connection. Among them, the graph model represents the association relationship between the parameter name and the parameter content information of the target item category, the node in the graph model represents the parameter name or parameter content information of the target item category, and the value corresponding to the edge in the graph model represents the association degree information between the parameter name and the parameter content information. Among them, the association degree information can be a value between 0 and 1.

[0040] It should be pointed out that the wireless connection mode can include but is not limited to 3G / 4G / 5G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other now known or future developed wireless connection modes.

[0041] Step 402, word segmentation is performed on the serialized text data to obtain a word set.

[0042] In some embodiments, the execution subject can perform word segmentation on the serialized text data to obtain a word set. As an example, the serialized text data can be segmented by Jieba word segmentation to obtain a word set.

[0043] Step 403, matching the target word in the word set with the parameter name in the graph model and generating a matched graph model according to the matching result.

[0044] In some embodiments, the execution subject can match the target word in the word set with the parameter name in the graph model and generate a matched graph model according to the matching result. Among them, the target word can be a word that has an association relationship with the parameter name in the graph model. As an example, the execution subject can convert the matching result into a graph form to obtain a matched graph model.

[0045] In some optional implementations of some embodiments, the matching of the target word in the word set with the parameter name in the graph model and the generation of the matched graph model according to the matching result can include the following steps:

[0046] In a first step, the target word in the word set is matched with the parameter names in the graph model to obtain a matching result. As an example, the execution subject can first extract semantic information of the target word in the word set. Then, by traversing each parameter name in the graph model, the parameter name related to the semantic information of the target word is determined, and finally, the parameter name related to the semantic information of the target word is matched with the target word to obtain a matching result.

[0047] In a second step, a matched graph model is generated according to the matching result and the target edge in the graph model. The target edge is an edge in the graph model representing the association degree information between the target word and the parameter name corresponding to the target word. As an example, the execution subject can determine the node information of the matched graph model according to the matching result. Finally, the target edge in the graph model is taken as an edge of the matched graph model to generate the matched graph model.

[0048] In step 404, a fused graph model is generated according to the structured text data and the matched graph model.

[0049] In some optional implementations of some embodiments, the execution subject can generate a fused graph model according to the structured text data and the matched graph model. As an example, the execution subject can integrate the structured text data into the matched graph model to obtain a fused graph model. Here, the fused graph model embodies the structured text information related to the target item.

[0050] In step 405, the summary information of the target item is generated according to the fused graph model.

[0051] In some embodiments, the execution subject can generate the summary information of the target item according to the fused graph model. As an example, the execution subject can first convert the fused graph model into a corresponding vector. Then, the corresponding vector and the vector corresponding to each node in the fused graph model are input into a pre-trained recurrent neural network to obtain the summary information of the target item.

[0052] In some optional implementations of some embodiments, the numerical value corresponding to the edge in the graph model is determined by the following steps:

[0053] In a first step, the item information of each item under the category of the target item is obtained, wherein the item information includes a parameter name and at least one parameter value.

[0054] Secondly, according to the item information of each item, the parameter content information corresponding to each parameter name and the number of occurrences of each parameter content information in the item information of each item are determined. Here, each item information in the item information of each item corresponds to at least one parameter name. Furthermore, the parameter name set corresponding to the item information of each item can be determined. Among them, there may be multiple repeated parameter names in the parameter name set. Therefore, the parameter name set can be de-duplicated to obtain the de-duplicated parameter name set. Here, each parameter name in the de-duplicated parameter name set corresponds to at least one parameter content information. Therefore, through the item information of each item, the number of occurrences of each parameter content information in the item information of each item can be determined.

[0055] Thirdly, according to the parameter content information corresponding to each parameter name and the number of occurrences of each parameter content information, the value corresponding to the edge in the graph model is determined by using the term frequency-inverse document frequency method.

[0056] Here, the main idea of the term frequency-inverse document frequency method is that if the frequency (Term Frequency, TF) of a word or phrase appearing in an article is high and rarely appears in other articles, it is considered that the word or phrase has good class distinguishing ability and is suitable for classification. While the inverse document frequency (Inverse Document Frequency, IDF) mainly refers to the fewer documents containing a word or phrase, the larger the IDF, which indicates that the word or phrase has good class distinguishing ability. Therefore, using the term frequency-inverse document frequency method, the importance of a word or phrase in an article can be calculated.

[0057] The above various embodiments of the present disclosure have the following beneficial effects: the information generation method of some embodiments of the present disclosure can accurately and effectively generate the summary information of the target item by considering the feature information of the serialized text data and the structured text data. Specifically, the recurrent neural network can only model the sequence text information, and since the item information includes not only unstructured text sequences but also structured information. The recurrent neural network is difficult to model directly for structured input, and the current processing method is still to process as a string sequence. This processing method may lose structured information, thereby reducing the expression ability of the model. Based on this, the information generation method of some embodiments of the present disclosure can first obtain a pre-constructed graph model associated with the target item category, serialized text data related to the target item, and structured text data. Among them, the graph model represents the association relationship between the parameter name and the parameter content information of the target item category, the nodes in the graph model represent the parameter name or the parameter content information of the target item category, 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 determine the parameter name corresponding to the target word in the subsequent word set, laying the foundation for subsequent generation of the matched graph model. It should be noted that the graph model is used to convert the serialized text data related to the target item into structured data. Then, the serialized text data is segmented to obtain a word set. Further, the target word in the word set and the parameter name in the graph model are matched, and a matched graph model is generated according to the matching result. Then, a fused graph model is generated according to the structured text data and the matched graph model. Here, the fused graph model embodies the structured text information related to the target item. Finally, the summary information of the target item is generated according to the fused graph model. Therefore, the information generation method can accurately and effectively generate the summary information of the target item by considering the feature information of the serialized text data and the structured text data.

[0058] With reference to Figure 5 , flow 500 illustrating another embodiment of the information generation method according to the present disclosure is shown. The information generation method comprises the following steps:

[0059] Step 501, obtaining a pre-constructed graph model associated with the target item category, serialized text data related to the target item, and structured text data.

[0060] Step 502, segmenting the serialized text data to obtain a word set.

[0061] Step 503, matching the target word in the word set and the parameter name in the graph model, and generating a matched graph model according to the matching result.

[0062] Step 504, generating a fused graph model according to the structured text data and the matched graph model.

[0063] In some embodiments, the specific implementation of steps 501-504 and the resulting technical effects can refer to Figure 4 The steps 401-404 in the corresponding embodiments are not described here.

[0064] Step 505, converting the fused graph model into a corresponding vector matrix.

[0065] In some embodiments, the execution subject (for example Figure 1 The electronic device 101 shown) can convert the fused graph model into a corresponding vector matrix. As an example, the execution subject can generate a corresponding vector matrix according to the numerical value corresponding to each edge in the fused graph model. Wherein the relationship between the isolated node and other nodes can be characterized as a numerical value of 0.

[0066] Step 506, determining the vector corresponding to each node in the fused graph model.

[0067] In some embodiments, the execution subject can determine the vector corresponding to each node in the fused graph model. Wherein the vector corresponding to each node represents the feature information of the parameter name or the parameter content information. As an example, the parameter name or the parameter content information corresponding to each node in the fused graph model can be converted into a corresponding vector by word embedding.

[0068] Step 507, 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.

[0069] In some embodiments, the execution subject can input the vector matrix and the vector corresponding to each node in the fused graph model into a pre-trained graph convolutional neural network (GCN, Graph Convolutional Network) to obtain a first output result. Wherein the graph convolutional network can be a multi-layer graph convolutional layer. The graph convolutional layer can aggregate the neighbor information of each node through an adjacency matrix, thereby enhancing the representation of the current node and improving the accuracy of the model for input modeling. Multi-layer graph convolutional layer modeling can obtain multi-hop neighbor information.

[0070] Step 508, encoding the first output result to obtain a second output result.

[0071] In some embodiments, the execution subject can encode the first output result to obtain a second output result. As an example, the execution subject can input the first output result into a pre-trained encoding network to obtain the second output result.

[0072] In some optional implementations of some embodiments, the execution subject can input the first output result into a pre-trained encoding network for processing time-series text data to obtain the second output result. The encoding network for processing time-series text data can be one of a recurrent neural network (RNN) and a long short-term memory (LSTM).

[0073] At step 509, the second output result is decoded to obtain the summary information of the target item.

[0074] In some embodiments, the execution subject can decode the second output result to obtain the summary information of the target item. As an example, the execution subject can input the second output result into a pre-trained decoding network to obtain the summary information of the target item.

[0075] In some optional implementations of some embodiments, the execution subject can input the second output result into a pre-trained decoding network for processing time-series text data to obtain the summary information of the target item. The decoding network for processing time-series text data can be one of a recurrent neural network and a long short-term memory.

[0076] As can be seen from Figure 5 compared with the description of some embodiments corresponding to Figure 4 compared with the description of some embodiments corresponding to Figure 5 The flow 500 of the information generation method in some embodiments corresponding to the above-mentioned fusion graph model embodies the step of generating the summary information of the target item according to the fusion graph model. Therefore, the schemes described in these embodiments can more accurately and effectively generate the summary information of the target item.

[0077] With reference to Figure 6 As an implementation of the above-mentioned methods, the present disclosure provides some embodiments of an information generation device. These device embodiments correspond to the above-mentioned method embodiments, and the device can be specifically applied to various electronic devices. Figure 4 The above-mentioned method embodiments correspond to the above-mentioned method embodiments, and the device can be specifically applied to various electronic devices.

[0078] As Figure 6As shown, the information generation apparatus 600 of some embodiments includes an acquisition unit 601, a word segmentation unit 602, a matching generation unit 603, a first generation unit 604, and a second generation unit 605. The acquisition unit 601 is configured to acquire a graph model associated with a target item category, serialized text data related to a target item, and structured text data, where the graph model represents the association between parameter names and parameter values of the target item category, the nodes in the graph model represent parameter names or parameter content information of the target item category, and the values corresponding to the edges in the graph model represent association degree information between the parameter names 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 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 first generation unit 604 is configured to generate a fused graph model based on the structured text data and the matched graph model. The second generation unit 605 is configured to generate summary information of the target item based on the fused graph model.

[0079] In some optional implementations of some embodiments, the matching generation unit 603 of the information generation apparatus 600 can be further configured to match target words in the word set with parameter names in the graph model to obtain a matching result, and generate a matched graph model based on the matching result and target edges in the graph model, where the target edges are edges in the graph model representing association degree information between the target words and parameter names corresponding to the target words.

[0080] In some optional implementations of some embodiments, the second generation unit 605 of the information generation apparatus 600 can be further configured to convert the fused graph model into a corresponding vector matrix, determine 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.

[0081] In some optional implementations of some embodiments, the second generation unit 605 of the information generation apparatus 600 can 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, encode the first output result to obtain a second output result, and decode the second output result to obtain the summary information of the target item.

[0082] In some optional implementations of some embodiments, the second generating unit 605 of the information generating apparatus 600 can be further configured to: input the first output result into a pre-trained encoding network for processing time-series text data to obtain the second output result.

[0083] In some optional implementations of some embodiments, the second generating unit 605 of the information generating apparatus 600 can be further configured to: input the second output result into a pre-trained decoding network for processing time-series text data to obtain the summary information of the target item.

[0084] It can be understood that the units described in the apparatus 600 correspond to the respective steps in the method described with reference to Figure 4 Thus, the operations, features, and advantages described above for the method also apply to the apparatus 600 and the units included therein, which will not be repeated here.

[0085] Reference is made to Figure 7 which shows a structural schematic diagram of an electronic device 700 suitable for implementing some embodiments of the present disclosure. Figure 7 The electronic device shown is merely an example and should not impose any limitation on the functions and use range of the embodiments of the present disclosure.

[0086] As shown in Figure 7 , the electronic device 700 can include a processing apparatus (e.g., a central processing unit, a graphics processing unit, etc.) 701, which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 702 or loaded from a storage apparatus 707 into a random access memory (RAM) 703. Various programs and data required for the operation of the electronic device 700 are also stored in the RAM 703. The processing apparatus 701, the ROM 702, and the RAM 703 are connected to each other through a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0087] Generally, the following apparatuses can be connected to the I / O interface 705: an input apparatus 706 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output apparatus 707 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage apparatus 708 including, for example, a magnetic tape, a hard disk, etc.; and a communication apparatus 709. The communication apparatus 709 can allow the electronic device 700 to communicate with other devices wirelessly or through wires to exchange data. Although Figure 7 The electronic device 700 is shown with various apparatuses, but it should be understood that it is not required to implement or have all of the shown apparatuses. More or fewer apparatuses can alternatively be implemented or included. Figure 7Each block in the flow diagrams represents one or more operations that can be implemented as hardware or software in embodiments. The representation is the functionally representative view of the operations but not the architectural representative view of the embodiments. In this context, each block signifies multiple entities, which can be combined or separated into several entities.

[0088] In particular, the processes described above with reference to the flowcharts can be implemented as computer software programs according to some embodiments of the present disclosure. For example, some embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer readable medium, the computer program comprising program code for performing the methods illustrated by 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, the above-mentioned functions defined in the methods of some embodiments of the present disclosure are performed.

[0089] It should be noted that the computer readable medium mentioned above in some embodiments of the present disclosure can be a computer readable signal medium or a computer readable storage medium or any combination of the two. The computer readable storage medium may, for example, be, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination of the above. More specific examples of the computer readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, 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 disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In some embodiments of the present disclosure, the computer readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus or device. In some embodiments of the present disclosure, the computer readable signal medium can include a data signal propagated in a baseband or as a carrier wave in a propagated data signal, which carries computer readable program code. Such propagated data signal can take many forms, including but not limited to, an electromagnetic signal, an optical signal or any suitable combination of the above. The computer readable signal medium can also be any computer readable medium that can send, propagate or transfer a program for use by or in connection with an instruction execution system, apparatus or device. The program code contained on the computer readable medium can be transmitted by any suitable medium, including but not limited to, wire, cable, RF (radio frequency), etc., or any suitable combination of the above.

[0090] In some embodiments, the client, server, or both can communicate using any known or future developed network protocols, such as the HyperText Transfer Protocol (HTTP), and can be interconnected with any form or medium of digital data communication (for example, a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), the Internet, and peer-to-peer networks (for example, ad hoc peer-to-peer networks), as well as any current or future developed network.

[0091] The computer-readable medium described above can be included in the device described above; or can exist alone, without being assembled into the electronic device. The computer-readable medium described above carries one or more programs, when the one or more programs are executed by the electronic device, the electronic device is caused to: acquire a pre-constructed graph model associated with a target article category, serialized text data and structured text data related to a target article, wherein the graph model represents an association relationship between a parameter name and each parameter content information of the target article category, a node in the graph model represents a parameter name or parameter content information of the target article category, and a value corresponding to an edge in the graph model represents 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 matched graph model according to a matching result; generate a fused graph model according to the structured text data and the matched graph model; and generate summary information of the target article according to the fused graph model.

[0092] Computer program code for carrying out operations of some embodiments of the present disclosure can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++, or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network ("LAN") or a wide area network ("WAN"), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0093] The computer program product of the first aspect can include a computer readable storage medium. The computer readable storage medium can include instructions. The instructions can include one or both of: instructions for causing a computer to implement a method as described above; and instructions for causing a computer to operate based on a system as described above. The computer readable storage medium can include one or more of: a magnetic disk; a magnetic disk drive; a magnetic tape; a magneto-optical drive; a solid state drive; a semiconductor drive; a flash drive; an optical drive; a holographic drive; a holographic medium; a memory stick; a floppy disk; a flexible disk; a hard disk; a hard disk drive; a holographic storage medium; a RAM drive; a ROM drive; a flash drive; an optical drive; a holographic drive; a magnetic drive; a removable memory port; a bay adapter; a memory PC card slot; a PC card socket; and a disk cartridge comprising a computer readable storage medium.

[0094] The units described in some embodiments of the present disclosure can be implemented by means of software, or by means of hardware. The described units can also be provided in a processor, for example, a processor can be described as comprising: an obtaining unit, a word segmentation unit, a matching generating unit, a first generating unit and a second generating unit. In some cases, the names of these units do not constitute a limitation on the units themselves, for example, the word segmentation unit can also be described as "a unit that segments the above-mentioned serialized text data to obtain a word set".

[0095] The functions described above in the text can be performed at least in part by one or more hardware logic components. For example, non-limiting example types of hardware logic components that can be used include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SOCs), complex programmable logic devices (CPLDs), and so on.

[0096] The above description is merely some of the preferred embodiments of the present disclosure and a description of the principles of the technology used. Those skilled in the art should understand that the scope of the application involved in the embodiments of the present disclosure is not limited to the technical solutions formed by the specific combinations of the above technical features, and should also cover other technical solutions formed by any combinations of the above technical features or equivalent features without departing from the above inventive concept. For example, the above features can be replaced with technical features disclosed in the embodiments of the present disclosure (but not limited to) having similar functions to form technical solutions.

Claims

1. A method for generating information, comprising: Obtain a pre-constructed graph model associated with the target item category, and obtain serialized text data and structured text data related to the target item. The graph model represents the association between the parameter names and the content information of each parameter of the target item category. The nodes in the graph model represent the parameter names or parameter content information of the target item category. The numerical values ​​corresponding to the edges in the graph model represent the degree of association between the parameter names and the parameter content information. The serialized text data is segmented to obtain a word set; Match the target words in the word set with the parameter names in the graph model, and generate the matched graph model based on the matching results; Based on the structured text data and the matched graph model, a fused graph model is generated; Based on the fused graph model, a summary of the target item is generated.

2. The method according to claim 1, wherein, The step of matching the target words in the word set with the parameter names in the graph model and generating the matched graph model based on the matching results includes: The target words in the word set are matched with the parameter names in the graph model to obtain the matching results; Based on the matching result and the target edge in the graph model, a matched graph model is generated, wherein the target edge is an edge in the graph model that represents the degree of association between the target word and the parameter name corresponding to the target word.

3. The method according to claim 1, wherein, The step of generating summary information of the target item based on the fused graph model includes: The fused graph model is then converted into a corresponding vector matrix. Determine the vector corresponding to each node in the fused graph model, wherein the vector corresponding to each node represents the feature information of the parameter name or the parameter content information; Based on the vector matrix and the vector corresponding to each node, a summary information of the target item is generated.

4. The method according to claim 3, wherein, The step of generating summary information for the target item based on the vector matrix and the vector corresponding to each node includes: 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 the first output result; The first output result is encoded to obtain the second output result; The second output result is decoded to obtain the summary information of the target item.

5. The method according to claim 4, wherein, Encoding the first output result to obtain the second output result includes: The first output result is input into a pre-trained encoding network that processes time-series text data to obtain the second output result.

6. The method according to claim 4, wherein, Decoding the second output result to obtain the summary information corresponding to the target item includes: The second output is input into a pre-trained decoding network that processes temporal text data to obtain the summary information of the target item.

7. The method according to claim 1, wherein, The numerical values ​​corresponding to the edges in the graph model are determined through the following steps: Obtain item information for each item under the target item category, wherein the item information includes a parameter name and at least one parameter value; Based on the item information of each item, determine the content information of each parameter corresponding to each parameter name and the number of times each parameter content information appears in the item information of each item; Based on the parameter content information corresponding to each parameter name and the frequency of occurrence of each parameter content information, the numerical values ​​corresponding to the edges in the graph model are determined using the term frequency-inverse file frequency method.

8. An information generation device, comprising: The acquisition unit is configured to acquire a pre-constructed graph model associated with the target item category, and to acquire serialized text data and structured text data related to the target item. The graph model represents the relationship between the parameter names and the values ​​of each parameter of the target item category. The nodes in the graph model represent the parameter names or parameter content information of the target item category. The values ​​corresponding to the edges in the graph model represent the degree of association between the parameter names and the parameter content information. The word segmentation unit is configured to segment the serialized text data into words to obtain a word set; The matching generation unit is configured to match the target words in the word set with the parameter names in the graph model and generate the matched graph model based on the matching results; The first generation unit is configured to generate a fused graph model based on the structured text data and the matched graph model. The second generation unit is configured to generate summary information of the target item based on the fused graph model.

9. An electronic device, comprising: One or more processors; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method as described in any one of claims 1-7.

10. A computer-readable medium having a computer program stored thereon, wherein, When the program is executed by the processor, it implements the method as described in any one of claims 1-7.

Images