Training apparatus for sns information integration model

Abstract

This invention relates to a training device for an SNS information integration model. The device includes: an information search device for searching a news database for the latest published personal social documents that match keywords input by the current user for news searching, serving as target news documents; the personal social documents storing single pieces of social information published by the user through the social networking service (SNS); and a model building device for building a feedforward neural network. The input layer data of the feedforward neural network consists of the personal social data corresponding to the single piece of social information published by the user through the SNS, and the output layer data consists of the search keywords tagged in the personal social documents. Through this invention, intelligent integration of various pieces of online information published by individuals can be achieved based on a feedforward neural network model, thereby enriching the news supply data while ensuring the timeliness of personally published information as pushed news content.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence, and more particularly to a training device for an SNS information integration model. Background Technology

[0002] SNS specifically refers to social networking services, including social software and social networking websites. It also refers to existing, mature, and widely used information carriers, such as SMS services. Another common interpretation of SNS is its full name: Social Networking Services, meaning "social networking website" or "social network." SNS can also refer to Social Network Software, a next-generation, personal network infrastructure software built using distributed technology, or more simply, P2P (Peer-to-Peer) technology.

[0003] In 1967, Stanley Milgram (1933–1984), a psychology professor at Harvard University, established the six degrees of separation theory. Simply put, "You are no more than six people away from any other stranger; that is, you can know any stranger through a maximum of six intermediaries." According to the six degrees of separation theory, each individual's social circle continuously expands, eventually forming a large network. This was an early understanding of social networking. Later, based on this theory, some people created internet services geared towards social networking, expanding social connections through "friends of friends," such as ArtComb, Friendster, Wallop, and Adomeme.

[0004] In the actual use of SNS, a large amount of personal online information includes news information with high real-time relevance. However, in the process of conversion and use, this news information requires tedious data reprocessing and special text format conversion by various news publishing terminals before it can be used as news material and enter the searchable state. This process requires a lot of manpower and material resources. The most disadvantageous aspect is that this operation mode greatly reduces the real-time nature of the news that users find. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a training device for an SNS information integration model. This device utilizes the characteristics of personally published online information to establish an artificial intelligence analysis model based on a feedforward neural network. It identifies and marks news search keywords for each piece of online information. When a user searches for news using the corresponding keywords, the device queries the latest matching personally published information based on the marked keywords and update time, and performs format conversion to obtain real-time news data from personally published online information on SNS.

[0006] Compared with the prior art, the present invention must possess at least the following three outstanding substantive features: (1) Build a feedforward neural network for performing news classification and aggregation of personal social data, and actively push subsequent news information based on the news classification and aggregation results; (2) The input layer data of the feedforward neural network is the personal social data corresponding to a single social information posted by an individual user through the social networking service SNS. The personal social data is binary data formed by concatenating the binary codes obtained by converting each character of the personal social document corresponding to the single social information into binary. The output layer data of the feedforward neural network is the search keywords marked on the personal social document. (3) The feedforward neural network is trained multiple times. Each training uses personal social data corresponding to personal social documents that have been manually labeled with search keywords as the input layer data of the feedforward neural network and manually labeled search keywords as the output layer data of the feedforward neural network. The number of training times is proportional to the character length of the input layer data to ensure the training effect of the feedforward neural network with input layer data of different complexity.

[0007] According to a first aspect of the present invention, a training apparatus for an SNS information integration model is provided, the apparatus comprising: An information search device is used to search a news database for the latest published personal social documents that match the keywords entered by the current user for searching news, and output them as target news documents. The personal social documents are used to store a single piece of social information published by an individual user through the social networking service SNS. A format conversion device, connected to the information search device, is used to convert the target news document output by the information search device into a text format that matches the browser currently selected by the user, so as to present it as a single news item in the browser; The model building device is used to build a feedforward neural network. The input layer data of the feedforward neural network is the personal social data corresponding to a single social information published by an individual user through a social networking service (SNS). The personal social data is binary data formed by concatenating the binary codes obtained by converting each character in the personal social document corresponding to the single social information into binary data. The output layer data of the feedforward neural network is the search keywords marked in the personal social document. A model learning device, connected to the model building device, is used to train the feedforward neural network multiple times. Each training session uses personal social data corresponding to personal social documents that have been manually labeled with search keywords as input layer data of the feedforward neural network and manually labeled search keywords as output layer data of the feedforward neural network. The number of training sessions is proportional to a set threshold. A data aggregation device, connected to the model learning device, is used to store the personal social document corresponding to each piece of social information, along with its identified search keywords and publication time, into the news database. The process of storing the personal social document corresponding to each piece of social information, along with its identified search keywords and publication time, into the news database includes: the identified search keywords are the search keywords obtained by the feedforward neural network after the corresponding personal social data is used as input layer data and trained multiple times. The number of characters undergoing binary conversion is fixed at a set threshold. The personal social data, which is binary data formed by concatenating the binary codes obtained by converting each character in the personal social document corresponding to the single social information into binary data, further includes: when the number of each character in the personal social document corresponding to the single social information is less than the set number threshold, the end of the corresponding personal social document is padded with zero characters to make up the difference for subsequent binary conversion. Specifically, storing the personal social document corresponding to each piece of social information, along with its identified search keywords and publication time, into the news database includes: the publication time of the personal social document corresponding to each piece of social information is the time corresponding to the first publication of the social information.

[0008] According to a second aspect of the present invention, a training method for an SNS information integration model is provided. The method includes using a training platform for an SNS information integration model as described above to classify and aggregate news data of social information published by individual users through social networking services based on a learned feedforward neural network model, so as to facilitate its subsequent use as news information actively pushed to users.

[0009] According to a third aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed, implements the steps of the training method for the SNS information integration model as described above. Attached Figure Description

[0010] The embodiments of the present invention will now be described with reference to the accompanying drawings, wherein: Figure 1This is a technical flowchart of the training device for the SNS information integration model according to the present invention.

[0011] Figure 2 This is a schematic diagram of the structure of a training device for an SNS information integration model according to Embodiment 1 of the present invention.

[0012] Figure 3 This is a schematic diagram of the structure of the training device for the SNS information integration model according to Embodiment 2 of the present invention.

[0013] Figure 4 This is a schematic diagram of the structure of the training device for the SNS information integration model according to Embodiment 3 of the present invention.

[0014] Figure 5 This is a schematic diagram of the structure of the training device for the SNS information integration model according to Embodiment 4 of the present invention.

[0015] Figure 6 This is a schematic diagram of a computer-readable storage medium according to Embodiment 6 of the present invention. Detailed Implementation

[0016] A social networking service is a platform that establishes connections between people, creating social networks or relationships. This includes, for example, sharing benefits, activities, backgrounds, or real-life connections. A social networking service includes social connections representing each user (usually a profile) and various additional services. Most social networking services are web-based online community services that provide users with means of interaction on the internet, such as email and instant messaging. Sometimes considered a single social networking service, but in a broader sense, a social networking service typically refers to a personal, community-centric service centered around online community groups. Social networking sites allow users to share their thoughts, pictures, articles, activities, and events on their networks.

[0017] In the process of news gathering, the personal news gathering model, with its wide distribution and large number of participants, possesses superior characteristics compared to the media gathering model, such as that of journalists. However, unfortunately, currently, when publishing news information obtained by individuals as news clients, it requires cumbersome procedures such as information collection, type analysis, format conversion, and even reprocessing of news content before the personal news information can be utilized. Obviously, this operational model is too complicated and wastes the immediacy of personal news gathering.

[0018] To overcome the above shortcomings, this invention establishes a training device for an SNS information integration model. First, a customized artificial intelligence analysis model based on a feedforward neural network is established for personally published online information. Then, historical online information with manually labeled search keywords is used to learn the artificial intelligence analysis model to obtain an intelligent parsing mechanism that can automatically parse the search keywords of each piece of personally published online information, thereby improving the utilization rate of online information from SNS individuals.

[0019] like Figure 1 The diagram shows a technical flowchart of a training apparatus for an SNS information integration model according to the present invention.

[0020] First, obtain each piece of personally posted online information from the cloud computing or big data service that executes the social networking service; Secondly, relevant personal social data and related time information are extracted from the information published online by individuals. Furthermore, a specially designed, learned feedforward neural network model is used to intelligently identify personal social data to obtain corresponding search keywords for news search, while also analyzing the first posting time of personal online information. Finally, the obtained search keywords and the first publication time are associated with the individual's published network information and stored in the news database, so that users can use news clients to conduct keyword-based news searches, thereby completing the classification and aggregation of SNS individual published network information as news push data.

[0021] The key point of this invention is that it abandons the previous classification and aggregation model of news push data for personal SNS-published network information. That is, when publishing news information obtained by individuals as a news client, it is necessary to go through cumbersome operation procedures such as information collection, type analysis, format conversion, and even news content reprocessing before the classification and aggregation model of personal news information can be utilized. Instead, it adopts a line-by-line calibration mechanism based on a learned feedforward neural network model to achieve intelligent integration of various network information published by individuals on SNS. This enriches the news supply data while ensuring the timeliness of personal information as push news content.

[0022] The training apparatus for the SNS information integration model of the present invention will now be described in detail by way of an embodiment.

[0023] Example 1 Figure 2 This is a schematic diagram of the structure of a training device for an SNS information integration model according to Embodiment 1 of the present invention.

[0024] like Figure 2As shown, the training device for the SNS information integration model includes the following components: An information search device is used to search a news database for the latest published personal social documents that match the keywords entered by the current user for searching news, and output them as target news documents. The personal social documents are used to store a single piece of social information published by an individual user through the social networking service SNS. A format conversion device, connected to the information search device, is used to convert the target news document output by the information search device into a text format that matches the browser currently selected by the user, so as to present it as a single news item in the browser; The model building device is used to build a feedforward neural network. The input layer data of the feedforward neural network is the personal social data corresponding to a single social information published by an individual user through a social networking service (SNS). The personal social data is binary data formed by concatenating the binary codes obtained by converting each character in the personal social document corresponding to the single social information into binary data. The output layer data of the feedforward neural network is the search keywords marked in the personal social document. A model learning device, connected to the model building device, is used to train the feedforward neural network multiple times. Each training session uses personal social data corresponding to personal social documents that have been manually labeled with search keywords as input layer data of the feedforward neural network and manually labeled search keywords as output layer data of the feedforward neural network. The number of training sessions is proportional to a set threshold. A data aggregation device, connected to the model learning device, is used to store the personal social document corresponding to each piece of social information, along with its identified search keywords and publication time, into the news database. The process of storing the personal social document corresponding to each piece of social information, along with its identified search keywords and publication time, into the news database includes: the identified search keywords are the search keywords obtained by the feedforward neural network after the corresponding personal social data is used as input layer data and trained multiple times. The number of characters undergoing binary conversion is fixed at a set threshold. The personal social data, which is binary data formed by concatenating the binary codes obtained by converting each character in the personal social document corresponding to the single social information into binary data, further includes: when the number of each character in the personal social document corresponding to the single social information is less than the set number threshold, the end of the corresponding personal social document is padded with zero characters to make up the difference for subsequent binary conversion. Specifically, storing the personal social document corresponding to each piece of social information, along with its identified search keywords and publication time, into the news database includes: the publication time of the personal social document corresponding to each piece of social information is the time corresponding to the first publication of the social information.

[0025] Example 2 Figure 3 This is a schematic diagram of the structure of the training device for the SNS information integration model according to Embodiment 2 of the present invention.

[0026] like Figure 3 As shown, compared to Embodiment 1 of the present invention, the training device for the SNS information integration model further includes: A timing service device, connected to the information search device, is used to provide timing services for the information search device to perform the action of searching a news database for the latest published personal social documents that match the keywords entered by the current user for searching news, and outputting them as target news documents. The timing service provided by the information search device for performing the action of searching the news database for the latest published personal social documents that match the keywords entered by the current user for searching news, and outputting them as target news documents, includes: the timing service device calculating the time difference between the current time and the publication time of each personal social document that matches the keywords entered by the current user for searching news, and outputting the personal social document with the smallest time difference as the target news document.

[0027] Example 3 Figure 4 This is a schematic diagram of the structure of the training device for the SNS information integration model according to Embodiment 3 of the present invention.

[0028] like Figure 4 As shown, compared to Embodiment 1 of the present invention, the training device for the SNS information integration model further includes: The user input device, connected to the information search device, is used to receive keywords entered by the current user through the browser they have selected for searching news.

[0029] Example 4 Figure 5 This is a schematic diagram of the structure of the training device for the SNS information integration model according to Embodiment 4 of the present invention.

[0030] like Figure 5 As shown, compared to Embodiment 1 of the present invention, the training device for the SNS information integration model further includes: A content storage device is connected to both the data aggregation device and the information search device, and is used to store the news database; The content storage device is also connected to the model building device and the model learning device, and is used to store the parameters of the feedforward neural network after each training session.

[0031] In any of the above embodiments, optionally, in the training device of the SNS information integration model: Converting the target news document output by the information search device into a text format matching the browser selected by the current user for presentation as a single news item in the browser includes: the text format matched by the browser selected by the current user is one of txt, html, and xml formats.

[0032] In any of the above embodiments, optionally, in the training device of the SNS information integration model: When the number of characters in the personal social data corresponding to the single social information is less than the set number threshold, the tail of the corresponding personal social data is padded with zero characters to make up the difference for subsequent binary conversion, including: the zero character is 0x00.

[0033] Example 5 In this embodiment, the present invention establishes a training method for an SNS information integration model. The method includes using the training platform of the SNS information integration model as described above to classify and aggregate news data of social information published by individual users through social networking services based on the learned feedforward neural network model so that it can be used as news information that is actively pushed later.

[0034] Example 6 Figure 6 This is a schematic diagram of a computer-readable storage medium according to Embodiment 6 of the present invention. Figure 6 As shown, the computer-readable storage medium 60 according to an embodiment of the present disclosure stores non-transitory computer-readable instructions 61 thereon. When the non-transitory computer-readable instructions 61 are executed by a processor, all or part of the steps of the training method for the SNS information integration model of Embodiment 5 of the present invention described above are performed.

[0035] Additionally, a feedforward neural network (FNN), or simply feedforward network, is a type of artificial neural network. A feedforward neural network employs a unidirectional, multi-layered structure. Each layer contains several neurons. In this type of neural network, each neuron receives signals from neurons in the previous layer and generates an output for the next layer. Layer 0 is called the input layer, the last layer is called the output layer, and the other intermediate layers are called hidden layers. Hidden layers can be single or multiple.

[0036] Common types of feedforward neural networks include perceptron networks, backpropagation (BP) networks, and RBF networks.

[0037] A perceptron (also called a perceptron) is the simplest feedforward network. It is mainly used for pattern classification, but can also be used in pattern classification-based learning control and multimodal control. Perceptron networks can be divided into single-layer perceptron networks and multi-layer perceptron networks.

[0038] A BP network is a feedforward network that uses the back propagation learning algorithm to adjust connection weights. Unlike a perceptron, the neuron transformation function of a BP network uses a sigmoid function, so the output is a continuous quantity between 0 and 1, which can realize arbitrary nonlinear mapping from input to output.

[0039] An RBF network is a feedforward network whose hidden layer neurons are composed of RBF neurons. An RBF neuron is a neuron whose transformation function is RBF (Radial Basis Function). A typical RBF network consists of three layers: an input layer, one or more RBF layers (hidden layers) composed of RBF neurons, and an output layer composed of linear neurons.

[0040] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0041] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software and a general-purpose hardware platform, or of course, using hardware. Based on this understanding, the above technical solutions, in essence or in terms of their contribution to related technologies, can be embodied in the form of software products.

[0042] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A training device for an SNS information integration model, characterized in that, The device includes: An information search device is used to search a news database for the latest published personal social documents that match the keywords entered by the current user for searching news, and output the searched personal social documents as target news documents. The personal social documents are used to store a single piece of social information published by an individual user through the social networking service SNS. A format conversion device, connected to the information search device, is used to convert the target news document output by the information search device into a text format that matches the browser currently selected by the user, so as to present it as a single news item in the browser; The model building device is used to build a feedforward neural network. The input layer data of the feedforward neural network is the personal social data corresponding to a single social information published by an individual user through a social networking service (SNS). The personal social data is binary data formed by concatenating the binary codes obtained by converting each character in the personal social document corresponding to the single social information into binary data. The output layer data of the feedforward neural network is the search keywords marked in the personal social document. A model learning device, connected to the model building device, is used to train the feedforward neural network multiple times. Each training session uses personal social data corresponding to personal social documents that have been manually labeled with search keywords as input layer data of the feedforward neural network, and uses manually labeled search keywords as output layer data of the feedforward neural network. The number of training sessions is proportional to a set threshold. A data aggregation device, connected to the model learning device, is used to store the personal social document corresponding to each piece of social information, along with its identified search keywords and publication time, into the news database. The process of storing the personal social document corresponding to each piece of social information, along with its identified search keywords and publication time, into the news database includes: the identified search keywords are the search keywords obtained by the feedforward neural network after the corresponding personal social data is used as input layer data and trained multiple times. The number of characters undergoing binary conversion is fixed at the set threshold. The personal social data, which is binary data formed by concatenating the binary codes obtained by converting each character in the personal social document corresponding to the single social information into binary data, further includes: when the number of each character in the personal social document corresponding to the single social information is less than the set number threshold, the end of the corresponding personal social document is padded with zero characters to make up the difference for subsequent binary conversion. Specifically, storing the personal social document corresponding to each piece of social information, along with its identified search keywords and publication time, into the news database includes: the publication time of the personal social document corresponding to each piece of social information is the time corresponding to the first publication of the social information.

2. The training device for the SNS information integration model as described in claim 1, characterized in that, The device further includes: A timing service device, connected to the information search device, provides timing services for the information search device to perform the action of searching a news database for the latest published personal social documents that match the keywords entered by the current user for searching news and outputting them as target news documents.

3. The training device for the SNS information integration model as described in claim 2, characterized in that: Providing a timing service for the information search device to perform the action of searching the news database for the latest published personal social documents that match the keywords entered by the current user for searching news and outputting them as target news documents includes: the timing service device calculates the time difference between the current time and the publication time of each personal social document searched from the news database that matches the keywords entered by the current user for searching news, and outputs the personal social document with the smallest time difference as the target news document.

4. The training device for the SNS information integration model as described in claim 1, characterized in that, The device further includes: The user input device, connected to the information search device, is used to receive keywords entered by the current user through the browser they have selected for searching news.

5. The training device for the SNS information integration model as described in claim 1, characterized in that, The device further includes: A content storage device is connected to both the data aggregation device and the information search device, and is used to store the news database.

6. The training device for the SNS information integration model as described in claim 5, characterized in that: The content storage device is also connected to the model building device and the model learning device, and is used to store the parameters of the feedforward neural network after each training session.

7. The training apparatus for the SNS information integration model as described in any one of claims 1 to 6, characterized in that: Converting the target news document output by the information search device into a text format that matches the browser currently selected by the user, so as to present it as a single news item in the browser, includes: the text format matched by the browser currently selected by the user is one of txt format, html format, and xml format.

8. The training apparatus for the SNS information integration model as described in any one of claims 1 to 6, characterized in that: When the number of characters in the personal social data stored in the personal social document corresponding to the single social information is less than the set number threshold, the end of the corresponding personal social document is padded with zero characters to make up the difference for subsequent binary conversion, including: the zero character being 0x00.

9. A training method for an SNS information integration model, characterized in that, The method includes using a training device for the SNS information integration model as described in any one of claims 1 to 8 to classify and aggregate news data of social information published by individual users through social networking services (SNS) based on a learned feedforward neural network model, so as to facilitate its subsequent use as news information that is actively pushed to users.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed, implements the steps of the method as described in claim 9.

Images