Information processing method and device, electronic equipment and storage medium

Abstract

The invention provides an information processing method and device, electronic equipment and a storage medium, and relates to the technical field of artificial intelligence, in particular to the technical fields of deep learning, large models, natural language processing and the like. According to the specific scheme, the method comprises the steps of obtaining text information, wherein the text information comprises first text information of a to-be-commented resource and second text information of a candidate prompt word prompt; according to the text information, selecting an optimal target prompt from the candidate prompts; and based on the to-be-commented resource and the target prompt, generating comment information of the to-be-commented resource. The optimal target prompt is selected from the candidate prompts, and the selected target prompt is most matched with the content of the to-be-evaluated resource, so that the generated comment information of the to-be-evaluated resource not only improves the quality of the comment information, but also can improve the accuracy and intelligent level of the comment information, and is beneficial to improving the user experience.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, specifically to technical fields such as deep learning, large models and natural language processing, and in particular to an information processing method, device, electronic device and storage medium. Background Art

[0002] On social media platforms, manual commenting is not only time-consuming and labor-intensive, but also inefficient and difficult to meet the needs of large-scale promotion and interaction. As the number of platform users continues to increase, more efficient ways are needed to manage their online image and interaction with users. With the continuous advancement of artificial intelligence technology, artificial intelligence algorithms can simulate human behavior and realize automated operations. For example, automatic comment tools, as an emerging comment tool, can help users automatically comment on resources on social media platforms. How to generate high-quality comments that match resources based on the prompts input by the automatic comment tool has become a hot topic of research. Summary of the invention

[0003] The present disclosure provides an information processing method, device, electronic device and storage medium.

[0004] According to one aspect of the present disclosure, there is provided an information processing method, comprising: acquiring text information, the text information comprising first text information of a resource to be reviewed and second text information of a candidate prompt word prompt;

[0005] Selecting an optimal target prompt from the candidate prompts according to the text information;

[0006] Based on the resource to be reviewed and the target prompt, review information of the resource to be reviewed is generated.

[0007] According to another aspect of the present disclosure, there is provided an information processing device, comprising:

[0008] An acquisition module, used to acquire text information, wherein the text information includes first text information of the resource to be reviewed and second text information of the candidate prompt word prompt;

[0009] A selection module, used to select an optimal target prompt from the candidate prompts according to the text information;

[0010] A generating module is used to generate comment information of the resource to be commented on based on the resource to be commented on and the target prompt.

[0011] According to another aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the information processing method described in the above-mentioned one aspect embodiment.

[0012] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, on which a computer program / instructions are stored, and the computer instructions are used to enable the computer to execute the information processing method described in the above-mentioned embodiment.

[0013] According to another aspect of the present disclosure, a computer program product is provided, including a computer program / instruction, wherein when the computer program / instruction is executed by a processor, the information processing method described in the above-mentioned embodiment is implemented.

[0014] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0016] Figure 1 A flowchart of an information processing method provided by an embodiment of the present disclosure;

[0017] Figure 2 A flowchart of another information processing method provided by an embodiment of the present disclosure;

[0018] Figure 2a A schematic diagram of a text information combination provided by an embodiment of the present disclosure;

[0019] Figure 3 A flowchart of another information processing method provided by an embodiment of the present disclosure;

[0020] Figure 4 A flowchart of another information processing method provided by an embodiment of the present disclosure;

[0021] Figure 5 A schematic diagram of the structure of an information processing device provided by an embodiment of the present disclosure;

[0022] Figure 6 The present invention is a block diagram of an electronic device for implementing the information processing method according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0023] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0024] The information processing method, apparatus, and electronic device according to the embodiments of the present disclosure are described below with reference to the accompanying drawings.

[0025] Artificial Intelligence (AI) is a discipline that studies how computers can simulate certain thought processes and intelligent behaviors of human beings (such as learning, reasoning, thinking, planning, etc.). It includes both hardware-level technologies and software-level technologies. AI hardware technologies generally include computer vision technology, speech recognition technology, natural language processing technology, as well as learning / deep learning, big data processing technology, knowledge graph technology, and other aspects.

[0026] Natural Language Processing (NLP) is an important field in the fields of computer science and artificial intelligence. It studies various theories and methods that can achieve effective communication between humans and computers using natural language. Natural language processing is a science that integrates linguistics, computer science, and mathematics. Natural language processing is mainly used in machine translation, public opinion monitoring, automatic summarization, opinion extraction, text classification, question answering, text semantic comparison, speech recognition, etc.

[0027] Deep Learning (DL) is a new research direction in the field of Machine Learning (ML). It is introduced into machine learning to make it closer to its original goal - artificial intelligence. Deep learning is the internal laws and representation levels of learning sample data. The information obtained in the learning process is of great help in the interpretation of data such as text, images and sounds. Its ultimate goal is to enable machines to have analytical learning capabilities like humans and to recognize data such as text, images and sounds. Deep learning is a complex machine learning algorithm that has achieved results in speech and image recognition that far exceed previous related technologies.

[0028] Smart search is a new generation of search engine that combines artificial intelligence technology. In addition to providing traditional fast retrieval and relevance sorting functions, it can also provide user role registration, automatic identification of user interests, semantic understanding of content, intelligent information filtering and push, etc.

[0029] Large models refer to machine learning models with large-scale parameters and complex computational structures. These models are usually built with deep neural networks and have billions or even hundreds of billions of parameters. Large models are designed to improve the expressiveness and predictive performance of the model and to handle more complex tasks and data. Large models are widely used in various fields, including natural language processing, computer vision, speech recognition, and recommendation systems. Large models learn complex patterns and features by training massive amounts of data, have stronger generalization capabilities, and can make accurate predictions on unseen data.

[0030] Figure 1 A flowchart of an information processing method provided by an embodiment of the present disclosure.

[0031] like Figure 1 As shown, the information processing method may include:

[0032] S101, Figure 1 A flowchart of an information processing method provided by an embodiment of the present disclosure.

[0033] like Figure 1 As shown, the information processing method may include but is not limited to the following steps:

[0034] S101, obtaining text information, where the text information includes first text information of a resource to be reviewed and second text information of a candidate prompt.

[0035] In some embodiments, the resources to be commented on may be resources on various resource platforms, and the resource platforms may include but are not limited to: news clients, social media platforms, video sharing websites, etc. The resources to be commented on may include but are not limited to: articles, pictures, audio, and video. For example, an article may be a text material on an academic forum. For another example, it may be an image resource published on a social platform. For another example, the audio may be a song, music, radio drama, etc. For another example, the video may be a short video, or it may be a TV series, movie, or other resource.

[0036] In some embodiments, the resource to be commented on can be understood as a resource browsed by a user on a certain platform, for example, it can be a short video currently playing on a social platform.

[0037] In some embodiments, the first text information of the resource to be reviewed may be used to describe text information of the resource to be reviewed.

[0038] In some embodiments, the candidate prompt is a prompt word for generating a comment corresponding to the resource to be commented on. In the disclosed embodiment, there may be two or more candidate prompts. Optionally, multiple prompts may be directly input by the user, or multiple prompts may be generated based on the comment requirement information input by the user.

[0039] In some embodiments, the second text information of the candidate prompt may be generated based on the original content of the candidate prompt.

[0040] S102, selecting the best target prompt from the candidate prompts according to the text information.

[0041] In some embodiments, after acquiring the text information, the text information may be subjected to feature extraction, and based on the extracted features, the candidate prompts may be evaluated to determine the matching degree or relevance between the candidate prompts and the resource to be evaluated. Further, based on the matching degree or relevance of the candidate prompts, the most likely target prompt may be determined from the candidate prompts.

[0042] S103, generating review information of the resource to be reviewed based on the resource to be reviewed and the target prompt.

[0043] In some embodiments, after obtaining the target prompt, the review information of the resource to be reviewed can be generated based on the resource to be reviewed and the target prompt. Optionally, the resource to be reviewed and the target prompt are input into an automatic review tool, and the automatic review tool generates the review information of the resource to be reviewed.

[0044] In some embodiments, the review information of the resource to be reviewed may include, but is not limited to, at least one of: review text, review video, review picture, and review audio.

[0045] In some embodiments, the comment area corresponding to the user may include multiple comment filling areas, for example, may include but are not limited to a text filling area, a video filling area, a picture filling area and an audio filling area.

[0046] In some embodiments, after obtaining the review information of the resource to be reviewed, each type of review content in the review information may be filled into the corresponding filling area.

[0047] The information processing method provided in the embodiments of the present disclosure selects the optimal target prompt from the candidate prompts through the text information of the resource to be commented on and the candidate prompts, and the selected target prompt best matches the style of the resource to be commented on, so that the comment information of the generated resource to be commented on can fit the content of the resource to be commented on itself, which not only improves the quality of the comment information, but also improves the accuracy and intelligence level of the comment information, and is useful for increasing the user experience on the resource sharing platform.

[0048] Figure 2 A flowchart of an information processing method provided by an embodiment of the present disclosure.

[0049] like Figure 2 As shown, the information processing method may include but is not limited to the following steps:

[0050] S201, obtaining text information, where the text information includes first text information of a resource to be reviewed and second text information of a candidate prompt.

[0051] In some embodiments, after the first text information and the second text information are acquired, the first text information and the second text information may be combined to obtain text information.

[0052] In some embodiments, the first text information and the second text information can be combined according to the configured preset template to obtain text information, so that the model can accurately identify and analyze, and improve the processing efficiency and accuracy of the model. Optionally, the preset template may include a template identifier, a filling position for the first text information, and a filling position for the second text information, and further a spacer such as a spacer is provided between the filling position of the first text information and the filling position of the second text information. Figure 2a shown.

[0053] For example, the text information can be classified according to the relevant SUG: <ans>Resource title content <sug>The prompt content is constructed in the form of a prompt, where the prompt content retains the first 250 characters. Among them, the relevant SUG: <ans>It can be a template identifier, and the resource title content is the first text information. <sug>is a separator, and the prompt content is the second text information. It can be understood that this is only an example of a preset template and cannot be used as a condition to limit the present disclosure.

[0054] In some embodiments, the first text information may be key information of the resource to be reviewed, wherein the key information may include but is not limited to at least one of the subject, title, tag, and summary information. Determining the first text information by the subject, title, tag, and summary information of the resource to be reviewed can make the first text information more accurately represent the content of the resource to be reviewed.

[0055] Optionally, if the resource to be reviewed itself carries at least one of a title, a subject, and a tag, at least one of the title, the subject, and the tag may be directly used as the first text information of the resource to be reviewed.

[0056] Optionally, if the resource to be reviewed does not carry key information, such as any of the title, subject and tag, semantic analysis and content understanding can be performed on the resource to be reviewed, and first text information of the resource to be reviewed can be generated based on the understanding information of the resource to be reviewed.

[0057] Optionally, at least one of a title, a subject, and a tag of the resource to be reviewed may be generated based on the deep understanding information, and the at least one of the title, the subject, and the tag may be used as the first text information of the resource to be reviewed.

[0058] Optionally, summary information may be generated based on the in-depth understanding information of the resource to be reviewed, and the summary information may be used as the first text information of the resource to be reviewed.

[0059] Optionally, semantic analysis and content understanding of the resource to be reviewed can be performed based on a pre-trained large language model, and understanding information of the resource to be reviewed can be output.

[0060] In the disclosed embodiment, semantic analysis and information understanding of the resource to be reviewed can be performed, which can not only ensure the acquisition of the first text information, but also ensure that the first text information is more closely related to the resource to be reviewed, thereby truly reflecting the content of the resource.

[0061] In some embodiments, part of the content is intercepted from the original content in the candidate prompt as the second text information of the candidate prompt. Optionally, the front content can be intercepted from the original content of the candidate prompt as the second text information of the candidate prompt. Optionally, the content in the middle position can be intercepted from the original content of the candidate prompt as the candidate prompt. Optionally, the number of intercepted words can be set, and the original content of the candidate prompt is intercepted from front to back according to the number of intercepted words to obtain the second text information of the candidate prompt. Optionally, the number of intercepted words can be a pre-configured number of words, for example, it can be 250 words. By intercepting the main content of the prompt, the full information of the prompt is not required, thereby reducing the transportation volume, improving the prediction efficiency and saving computing resources.

[0062] In some embodiments, the original content of the candidate prompt may be summarized, and the summary information may be used as the second text information of the candidate prompt.

[0063] In some embodiments, the original content of the candidate prompt may be identified, and the content related to the user input information may be determined therefrom as the second text information of the candidate prompt.

[0064] In the disclosed embodiment, by acquiring the second text information from the prompt, it is ensured that the second text information can fully express the main information of the prompt, which is beneficial to the quality prediction of the subsequent prompt.

[0065] The optional implementation of step S201 can be found in Figure 1 Optional implementations of step S101, and Figure 1 Other related parts in the embodiments involved will not be described in detail here.

[0066] S202: predicting quality parameters of the candidate prompts based on the text information.

[0067] In some embodiments, the text information may be segmented and sparse features may be extracted to obtain a segmentation identification (ID) of the text information and an ID of the sparse feature. Further, the segmentation ID and the sparse feature ID may be fused to obtain a fused feature.

[0068] Furthermore, quality parameters of the candidate prompt are predicted based on the fusion features to obtain the predicted quality parameters of the candidate prompt.

[0069] In some embodiments, the quality parameters may include, but are not limited to: the score of the candidate prompt, the matching degree, and the dwell time of the comment.

[0070] Optionally, the score of the candidate prompt can be used to evaluate the quality of the candidate prompt. Optionally, the matching degree of the candidate prompt can be used to evaluate the matching degree or relevance degree between the candidate prompt and the resource to be evaluated.

[0071] Optionally, the dwell time of the comment of the candidate prompt can be used to reflect the length of time the user stays on the comment information generated based on the candidate prompt, and further, can reflect the quality of the candidate prompt. In the disclosed embodiment, the dwell time of the comment information is positively correlated with the quality of the comment information, that is, the longer the dwell time of the comment information, the better the quality of the comment information, and the shorter the dwell time of the comment information, the worse the quality of the comment information.

[0072] It is understandable that the shorter the retention time of the comment information, the less attractive the comment information is to the user, or the user has a low interest in the comment information, and the user's willingness to interact with the comment information is low, which cannot effectively achieve the retention of users. On the other hand, the longer the retention time of the comment information, the more attractive the comment information is to the user, or the user has a high interest in the comment information, which is conducive to the user's interaction with the comment information, such as commenting on the comment information, liking it, etc., which can reduce the possibility of user loss and provide user retention.

[0073] S203: Select an optimal target prompt from the candidate prompts according to the quality parameters of the candidate prompts.

[0074] In some embodiments, the candidate prompts may be ranked based on their quality parameters, and the prompt with the best quality may be selected as the target prompt.

[0075] In some embodiments, the candidate prompts may be ranked from high to low based on their scores, and the candidate prompt ranked at the top may be selected as the target prompt with the best quality.

[0076] In some embodiments, the candidate prompts may be sorted from high to low based on their matching degrees, and the candidate prompt at the head of the sort may be selected as the target prompt with the best quality.

[0077] In some embodiments, the candidate prompts can be sorted from long to short based on the length of time the comments of the candidate prompts remain, and the candidate prompt at the top of the sorting can be selected as the target prompt with the best quality to increase the user's interest in the comment information, which is conducive to improving the user experience.

[0078] In the disclosed embodiment, the optimal target prompt is selected from the candidate prompts, and the selected target prompt best matches the style of the resource to be evaluated, so that high-quality review information that is more in line with the content of the resource to be evaluated can be generated, allowing more users to see richer review content, thereby increasing the push of reviews.

[0079] S204: Generate review information of the resource to be reviewed based on the resource to be reviewed and the target prompt.

[0080] Optionally, it can be a combination of comment text and comment video. Optionally, it can be a combination of comment text and comment picture. Optionally, it can be a combination of comment text and comment audio. In the disclosed embodiment, multimodal comment information can be generated, which is conducive to enriching the form of comments, further increasing the user's interest in the comment information and improving the user experience.

[0081] In some embodiments, the comment video may be generated based on the comment text, or may be a related video searched from a video library based on the comment text, or may be a related video searched from a video library based on the resource to be commented on. Optionally, key information and tag information of the resource to be commented on may be obtained, and related videos may be searched from a video library based on the key information and tag information.

[0082] The optional implementation of step S204 can be found in Figure 1 Optional implementations of steps S104 and S103, and Figure 1 Other related parts in the embodiments involved will not be described in detail here.

[0083] The information processing method provided in the disclosed embodiment selects the optimal target prompt from the candidate prompts through the text information of the resource to be commented on and the candidate prompts, and the selected target prompt best matches the style of the resource to be evaluated, so that the comment information of the generated resource to be evaluated can fit the content of the resource to be evaluated itself, which not only improves the quality of the comment information, but also improves the accuracy and intelligence level of the comment information, and is useful for improving the user experience of the resource sharing platform. Especially on social media platforms, the number of comments is often related to the popularity of resources. Intelligent comments can simplify comment operations, increase comment speed, and quickly increase the number of comments, thereby making it easier to push resources to be rated. Through intelligent comment drainage, more users can pay attention to the resources to be evaluated. At the same time, intelligent comments can also be used to handle user inquiries and feedback, which can solve the problem of untimely feedback, thereby further improving user satisfaction and experience.

[0084] Figure 3 A flowchart of an information processing method provided by an embodiment of the present disclosure.

[0085] like Figure 3 As shown, the information processing method may include but is not limited to the following steps:

[0086] S301, obtaining text information, where the text information includes first text information of a resource to be reviewed and second text information of a candidate prompt.

[0087] The optional implementation of step S301 can be found in Figure 1 Step S101 and Figure 2 Optional implementations of step S201, and Figure 1 and Figure 2 Other related parts in the embodiments involved will not be described in detail here.

[0088] S302, performing word segmentation on the text information, and obtaining a coding vector of the text information according to the word segmentation.

[0089] In some embodiments, text information is segmented by a tokenizer, and the tokens are processed by a first embedding layer to obtain representation vectors of the tokens. Further, the representation vectors of the tokens are encoded by an encoder to obtain encoding vectors of the text information, wherein the encoding vectors of the text information include token ID information of the tokens.

[0090] S303, extracting sparse features from the text information and obtaining a representation vector of the sparse features.

[0091] In some embodiments, the sparse features that can be extracted from the text information may include, but are not limited to: prompt number, resource type, resource category, resource interest point, resource cover image classification, multiple combined features of the above information, prompt type, and multiple combined features of partial information in the above information. Optionally, the sparse features are shown in Table 1 below:

[0092] Table 1 Sparse feature table

[0093] Sparse features Description prompt_id prompt number vt Resource type, news / mv / sv cate Resource Level 1 Category subcate Resource secondary category att_0 Resource Point of Interest 1 att_1 Resource Point of Interest 2 att_2 Resource Point of Interest 3 cover_cate Resource cover image classification prompt_vt Combined features, prompt_id-vt prompt_cate Combined features, prompt_id-first level category prompt_subcate Combined features, prompt_id-secondary category prompt_cover_cate Combination features, prompt_id-cover image classification prompt_type prompt type, rag / deep prompt_type_cate Combined features, prompt type - primary category prompt_type_subcate Combined features, prompt type-secondary category

[0094] It is understood that each element in Table 1 exists independently. These elements are exemplarily listed in the same table, but it does not mean that all elements in the table must exist at the same time as shown in the table. The value of each element is independent of the value of any other element in Table 1. Therefore, those skilled in the art can understand that the value of each element in Table 1 is an independent embodiment.

[0095] In some embodiments, sparse features may be extracted based on the above table 1 to obtain IDs of coefficient features, and further, the IDs of the sparse features may be processed through a second embedding layer to obtain representation vectors of the sparse features.

[0096] S304: predicting the quality parameters of the candidate prompts according to the encoding vector of the text information and the representation vector of the sparse features.

[0097] In some embodiments, after obtaining the encoding vector of the text information and the representation vector of the sparse feature, the encoding vector and the sparse feature of the text information can be spliced ​​to obtain a spliced ​​vector. Further, the spliced ​​vector is input into a deep neural network (DNN) based on a multilayer perceptron (MLP) for quality prediction, and the DNN predicts the quality parameters of the candidate prompt based on the fusion vector.

[0098] In the present embodiment, in the process of predicting the quality parameters of the candidate prompts, both the text features and the sparse features of the text information are taken into consideration, thereby improving the prediction accuracy of the prompt's quality parameters, facilitating the subsequent selection of an accurate target prompt, and thereby ensuring the quality of the generated comment information.

[0099] S305: Select an optimal target prompt from the candidate prompts according to the quality parameters of the candidate prompts.

[0100] The optional implementation of step S305 can be found in Figure 1 Step S102 and Figure 2 Optional implementations of step S203, and Figure 1 and Figure 2 Other related parts in the embodiments involved will not be described in detail here.

[0101] S306: Generate review information of the resource to be reviewed based on the resource to be reviewed and the target prompt.

[0102] The optional implementation of step S306 can be found in Figure 1 Step S103 and Figure 2 Optional implementations of step S204, and Figure 1 and Figure 2 Other related parts in the embodiments involved will not be described in detail here.

[0103] The information processing method provided in the embodiments of the present disclosure selects the optimal target prompt from the candidate prompts through the text information of the resource to be commented on and the candidate prompts, and the selected target prompt best matches the style of the resource to be commented on, so that the comment information of the generated resource to be commented on can fit the content of the resource to be commented on itself, which not only improves the quality of the comment information, but also improves the accuracy and intelligence level of the comment information, and is helpful in increasing the user experience of the resource sharing platform.

[0104] Figure 4 An information processing model provided by the embodiment of the present disclosure, such as Figure 4 As shown, the information processing model can be a pre-trained multilingual text-to-text transfer transformer (MT5) model and DNN. In the present disclosure, text information can be input into the MT5 model, and the MT5 model can segment the text information to obtain multiple segmentations (tokens), and the tokens can be further encoded to obtain the segmentation ID of the token, and the ID of the sparse features of the text information can be obtained. Furthermore, based on the segmentation ID of the token and the ID of the sparse features, the candidate prompt is predicted to obtain the predicted label of the candidate prompt, which can be the comment residence time. Optionally, the comment residence time can be a normalized value in the range of (0 to 1).

[0105] In some embodiments, the MT5 model may include an input layer, a tokenizer, a first embedding layer, an encoder, a sparse feature layer, and a second embedding layer. The input layer, the tokenizer, the first embedding layer, and the encoder are connected in sequence, the input layer is connected to the sparse feature layer, the sparse feature layer is connected to the second embedding layer, and the encoder and the second embedding layer are both connected to the DNN.

[0106] In some embodiments, the encoder can be a self-attention mechanism Transformer structure.

[0107] In this embodiment, the input layer inputs the text information into the Tokenizer, which segments the text information to obtain multiple tokens. Further, the token output by the Tokenizer is input into the first Embedding Layer, and the token is processed by the first Embedding Layer to obtain a representation vector of the token. Further, the representation vector of the token is input into the encoder for encoding to obtain an encoding vector of the text information, which carries the segmentation ID information. Further, the input layer inputs the text information into the sparse feature layer, and the sparse feature layer extracts sparse features, and the ID of the sparse feature is input into the second Embedding Layer to process the sparse features to obtain a representation vector of the sparse features.

[0108] Furthermore, the encoding vector of the text information and the representation vector of the sparse features are input into the DNN, and the DNN outputs the predicted label of the candidate prompt's evaluation residence time based on the encoding vector of the text information and the representation vector of the sparse features. Optionally, the concatenated vector is reduced in dimension and input into a classification head composed of multiple layers of fully connected layers and corresponding activation functions, and finally outputs a score estimate between 0 and 1, which is used to represent the comment residence time. The larger the score, the longer the residence time, and the smaller the score, the shorter the residence time.

[0109] The information processing method provided in the disclosed embodiment selects the optimal target prompt from the candidate prompts through the text information of the resource to be commented on and the candidate prompts, and the selected target prompt is most matched with the style of the resource to be evaluated, so that the comment information of the generated resource to be evaluated can fit the content of the resource to be evaluated itself, which not only improves the quality of the comment information and allows more users to see richer comment content, but also improves the accuracy and intelligence level of the comment information, and is useful for improving the user experience of the resource sharing platform. Especially on social media platforms, the number of comments is often related to the popularity of resources. Intelligent comments can simplify comment operations, increase comment speed, and quickly increase the number of comments, thereby making it easier to push resources to be rated. Through intelligent comment drainage, more users can pay attention to the resources to be evaluated. At the same time, intelligent comments can also be used to handle user inquiries and feedback, which can solve the problem of untimely feedback, thereby further improving user satisfaction and experience.

[0110] Corresponding to the information processing methods provided in the above-mentioned embodiments, an embodiment of the present disclosure further provides an information processing device. Since the information processing device provided in the embodiment of the present disclosure corresponds to the information processing methods provided in the above-mentioned embodiments, the implementation methods of the above-mentioned information processing methods are also applicable to the information processing device provided in the embodiment of the present disclosure and will not be described in detail in the following embodiments.

[0111] Figure 5 A schematic diagram of the structure of an information processing device provided in an embodiment of the present disclosure.

[0112] like Figure 5 As shown, the information processing device 500 of the embodiment of the present disclosure includes: an acquisition module 501, a selection module 502 and a generation module 503.

[0113] An acquisition module 501 is used to acquire text information, wherein the text information includes first text information of a resource to be reviewed and second text information of a candidate prompt word prompt;

[0114] A selection module 502, configured to select an optimal target prompt from the candidate prompts according to the text information;

[0115] The generating module 503 is used to generate the review information of the resource to be reviewed based on the resource to be reviewed and the target prompt.

[0116] In some embodiments, the selection module 502 is further configured to:

[0117] Predicting quality parameters of the candidate prompt according to the text information;

[0118] An optimal target prompt is selected from the candidate prompts according to the quality parameters of the candidate prompts.

[0119] In some embodiments, the quality parameter is the comment retention time, and the selection module 502 is further used to:

[0120] The candidate prompts are sorted according to the comment residence time, and the candidate prompt with the longest comment residence time is selected as the target prompt.

[0121] In some embodiments, the selection module 502 is further configured to:

[0122] Segmenting the text information, and obtaining an encoding vector of the text information according to the segmentation;

[0123] Extracting sparse features from the text information and obtaining a representation vector of the sparse features;

[0124] The quality parameter of the candidate prompt is predicted according to the encoding vector and the representation vector of the sparse feature.

[0125] In some embodiments, the selection module 502 is further configured to:

[0126] Segmenting the text information by a word segmenter, and processing the segmented words by a first embedding layer to obtain a representation vector of the segmented words;

[0127] The representation vector of the word segment is encoded by an encoder to obtain an encoding vector of the text information, where the encoding vector carries information of the word segment ID of the word segment.

[0128] In some embodiments, the selection module 502 is further configured to:

[0129] The ID of the sparse feature is obtained, and the ID of the sparse feature is processed through a second embedding layer to obtain a vector representation of the sparse feature.

[0130] In some embodiments, the selection module 502 is further configured to:

[0131] Concatenating the encoding vector and the representation vector of the sparse feature to obtain a concatenated vector;

[0132] The splicing vector is input into a deep neural network DNN based on a multi-layer perceptron MLP, and the DNN predicts the quality parameter of the candidate prompt based on the splicing vector.

[0133] In some embodiments, the acquisition module 501 is further used to:

[0134] At least one of the key information of the resource to be reviewed is obtained, and the first text information is generated according to the at least one of the key information, wherein the key information at least includes a title, a subject, and a summary information.

[0135] In some embodiments, the acquisition module 501 is further used to:

[0136] In response to the resource to be reviewed not carrying at least one of the key information, semantic analysis and content understanding are performed on the resource to be reviewed, and the first text information is generated based on the understanding information.

[0137] In some embodiments, the acquisition module 501 is further used to:

[0138] The original content of the candidate prompt is obtained, and the second text information is intercepted from the original content of the candidate prompt.

[0139] In some embodiments, the acquisition module 501 is further used to:

[0140] The number of words to be cut is determined, and the original content of the candidate prompt is cut from front to back based on the number of words to obtain the second text information.

[0141] In some embodiments, the acquisition module 501 is further used to:

[0142] The first text information and the second text information are combined according to a preset template to obtain the text information.

[0143] In some embodiments, the review information of the resource to be reviewed includes at least one of a review text, a review picture, a review audio, and a review video of the resource to be reviewed.

[0144] The information processing method provided in the disclosed embodiment selects the optimal target prompt from the candidate prompts through the text information of the resource to be commented on and the candidate prompts, and the selected target prompt is most matched with the style of the resource to be evaluated, so that the comment information of the generated resource to be evaluated can fit the content of the resource to be evaluated itself, which not only improves the quality of the comment information and allows more users to see richer comment content, but also improves the accuracy and intelligence level of the comment information, and is useful for improving the user experience of the resource sharing platform. Especially on social media platforms, the number of comments is often related to the popularity of resources. Intelligent comments can simplify comment operations, increase comment speed, and quickly increase the number of comments, thereby making it easier to push resources to be rated. Through intelligent comment drainage, more users can pay attention to the resources to be evaluated. At the same time, intelligent comments can also be used to handle user inquiries and feedback, which can solve the problem of untimely feedback, thereby further improving user satisfaction and experience.

[0145] In the technical solution disclosed herein, the acquisition, storage and application of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0146] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0147] Figure 6 A schematic block diagram of an example electronic device 600 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0148] like Figure 6 As shown, the device 600 includes a computing unit 601, which can perform various appropriate actions and processes according to a computer program / instruction stored in a read-only memory (ROM) 602 or a computer program / instruction loaded from a storage unit 606 to a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the device 600 can also be stored. The computing unit 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0149] A number of components in the device 600 are connected to the I / O interface 605, including: an input unit 606 such as a keyboard, a mouse, etc.; an output unit 607 such as various types of displays, speakers, etc.; a storage unit 608 such as a disk, an optical disk, etc.; and a communication unit 609 such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 609 allows the device 600 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0150] The computing unit 601 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 601 performs the various methods and processes described above, such as information processing methods. For example, in some embodiments, the information processing method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 606. In some embodiments, part or all of the computer program / instructions may be loaded and / or installed on the device 600 via ROM 602 and / or communication unit 609. When the computer program / instructions are loaded into RAM 603 and executed by the computing unit 601, one or more steps of the information processing method described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the information processing method in any other appropriate manner (e.g., by means of firmware).

[0151] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs / instructions that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0152] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0153] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, 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 suitable combination of the foregoing.

[0154] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0155] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), the Internet, and a blockchain network.

[0156] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs / instructions running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0157] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in the disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in the disclosure can be achieved, and this document does not limit them here.

[0158] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.< / sug> < / ans> < / sug> < / ans>

Claims

1. An information processing method, wherein: The method comprises: Acquire text information, the text information including first text information of the resource to be reviewed and second text information of the candidate prompt word prompt; Selecting an optimal target prompt from the candidate prompts according to the text information; Based on the resource to be reviewed and the target prompt, review information of the resource to be reviewed is generated.

2. The method according to claim 1, wherein: The selecting the optimal target prompt from the candidate prompts according to the text information includes: Predicting quality parameters of the candidate prompt according to the text information; An optimal target prompt is selected from the candidate prompts according to the quality parameters of the candidate prompts.

3. The method according to claim 2, wherein: The quality parameter is the dwell time of the comment, and selecting the optimal target prompt from the candidate prompts according to the quality parameter of the candidate prompts includes: The candidate prompts are sorted according to the comment residence time, and the candidate prompt with the longest comment residence time is selected as the target prompt.

4. The method according to claim 2, wherein: The predicting, according to the text information, the quality parameter of the candidate prompt includes: Segmenting the text information, and obtaining an encoding vector of the text information according to the segmentation; Extracting sparse features from the text information and obtaining a representation vector of the sparse features; The quality parameter of the candidate prompt is predicted according to the encoding vector and the representation vector of the sparse feature.

5. The method according to claim 4, wherein: The word segmentation of the text information and obtaining the encoding vector of the text information according to the word segmentation includes: Segmenting the text information by a word segmenter, and processing the segmented words by a first embedding layer to obtain a representation vector of the segmented words; The representation vector of the word segment is encoded by an encoder to obtain an encoding vector of the text information, where the encoding vector carries information of the word segment ID of the word segment.

6. The method according to claim 4, wherein: The obtaining of the representation vector of the sparse feature comprises: The ID of the sparse feature is obtained, and the ID of the sparse feature is processed through a second embedding layer to obtain a vector representation of the sparse feature.

7. The method according to claim 4, wherein: The predicting the quality parameter of the candidate prompt according to the encoding vector and the representation vector of the sparse feature includes: Concatenating the encoding vector and the representation vector of the sparse feature to obtain a concatenated vector; The splicing vector is input into a deep neural network DNN based on a multi-layer perceptron MLP, and the DNN predicts the quality parameter of the candidate prompt based on the splicing vector.

8. The method according to any one of claims 1 to 7, wherein: The process of obtaining the first text information includes: At least one of the key information of the resource to be reviewed is obtained, and the first text information is generated according to the at least one of the key information, wherein the key information at least includes a title, a subject, and a summary information.

9. The method according to claim 8, wherein: The process of obtaining the first text information includes: In response to the resource to be reviewed not carrying at least one of the key information, semantic analysis and content understanding are performed on the resource to be reviewed, and the first text information is generated based on the understanding information.

10. The method according to claim 1, wherein: The process of obtaining the second text information includes: The original content of the candidate prompt is obtained, and the second text information is intercepted from the original content of the candidate prompt.

11. The method according to claim 10, wherein: The step of intercepting the second text information from the original content of the candidate prompt includes: The number of words to be cut is determined, and the original content of the candidate prompt is cut from front to back based on the number of words to obtain the second text information.

12. The method according to any one of claims 1 to 7, wherein: The method further comprises: The first text information and the second text information are combined according to a preset template to obtain the text information.

13. The method according to any one of claims 1 to 7, wherein: The review information of the resource to be reviewed includes at least one of a review text, a review picture, a review audio, and a review video of the resource to be reviewed.

14. An information processing device, wherein: The device comprises: An acquisition module, used to acquire text information, wherein the text information includes first text information of the resource to be reviewed and second text information of the candidate prompt word prompt; A selection module, used to select an optimal target prompt from the candidate prompts according to the text information; A generating module is used to generate comment information of the resource to be commented on based on the resource to be commented on and the target prompt.

15. The device according to claim 14, wherein: The selection module is further used for: Predicting quality parameters of the candidate prompt according to the text information; An optimal target prompt is selected from the candidate prompts according to the quality parameters of the candidate prompts.

16. The device according to claim 15, wherein: The quality parameter is the dwell time of the comment. The selection module is further used for: The candidate prompts are sorted according to the comment residence time, and the candidate prompt with the longest comment residence time is selected as the target prompt.

17. The device according to claim 15, wherein: The selection module is also used for: Segmenting the text information, and obtaining an encoding vector of the text information according to the segmentation; Extracting sparse features from the text information and obtaining a representation vector of the sparse features; The quality parameter of the candidate prompt is predicted according to the encoding vector and the representation vector of the sparse feature.

18. The device according to claim 17, wherein: The selection module is also used for: Segmenting the text information by a word segmenter, and processing the segmented words by a first embedding layer to obtain a representation vector of the segmented words; The representation vector of the word segment is encoded by an encoder to obtain an encoding vector of the text information, where the encoding vector carries information of the word segment ID of the word segment.

19. The device according to claim 17, wherein: The selection module is also used for: The ID of the sparse feature is obtained, and the ID of the sparse feature is processed through a second embedding layer to obtain a vector representation of the sparse feature.

20. The device according to claim 17, wherein: The selection module is further used for: Concatenating the encoding vector and the representation vector of the sparse feature to obtain a concatenated vector; The splicing vector is input into a deep neural network DNN based on a multi-layer perceptron MLP, and the DNN predicts the quality parameter of the candidate prompt based on the splicing vector.

21. The device according to any one of claims 14 to 20, wherein: The acquisition module is further used for: At least one of the key information of the resource to be reviewed is obtained, and the first text information is generated according to the at least one of the key information, wherein the key information at least includes a title, a subject, and a summary information.

22. The device according to claim 21, wherein The acquisition module is further used for: In response to the resource to be reviewed not carrying at least one of the key information, semantic analysis and content understanding are performed on the resource to be reviewed, and the first text information is generated based on the understanding information.

23. The device according to claim 14, wherein: The acquisition module is further used for: The original content of the candidate prompt is obtained, and the second text information is intercepted from the original content of the candidate prompt.

24. The device according to claim 23, wherein: The acquisition module is further used for: The number of words to be cut is determined, and the original content of the candidate prompt is cut from front to back based on the number of words to obtain the second text information.

25. The device according to any one of claims 14 to 27, wherein: The acquisition module is further used for: The first text information and the second text information are combined according to a preset template to obtain the text information.

26. The device according to any one of claims 14 to 27, wherein: The review information of the resource to be reviewed includes at least one of a review text, a review picture, a review audio, and a review video of the resource to be reviewed.

27. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 13.

28. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-13.

29. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the method steps of any one of claims 1-13 are implemented.

Images