Information processing method and device, equipment and storage medium

Abstract

The embodiment of the invention relates to an information processing method and device, equipment and a storage medium. The method comprises the steps of obtaining a to-be-processed target task and initial association information; the initial association information is stored by taking the first statement granularity as a unit; splitting the initial associated information to generate a plurality of pieces of first associated information of a second statement granularity; the second statement granularity is smaller than the first statement granularity; the first associated information is information with complete semantics; performing feature extraction on the target problem and each piece of first associated information, and determining the association degree between the extracted features as the association degree between the target task and each piece of first associated information; screening second associated information of which the association degree is greater than a first threshold value from the first associated information; and performing content summarization information extraction processing on the second associated information to generate target associated information. Therefore, the data volume of the target associated information and the loss probability of the key information are reduced, and the accuracy of the target associated information is improved.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to an information processing method, apparatus, device and storage medium. Background Art

[0002] With the development of network technology, many services are now implemented through distributed deployment. During the operation of the service, different computing nodes need to transfer and share information with related relationships (referred to as related information or context information) to effectively collaborate and jointly complete complex tasks. The transfer of such related information has an important impact on the overall performance and success rate of the distributed system.

[0003] At present, there are two main ways to implement the transmission of related information: one is to transmit all the related information; the other is to summarize all the related information and transmit the summarized related information. However, the first method is prone to cause a significant increase in computing costs and communication overheads due to the large amount of data transmitted; the second method is prone to cause the loss of important information, affecting the accuracy and efficiency of the system response. Summary of the invention

[0004] In order to solve the above technical problems, the present disclosure provides an information processing method, apparatus, device and storage medium.

[0005] In a first aspect, an embodiment of the present disclosure provides an information processing method, the method comprising:

[0006] Acquire the target task to be processed and initial associated information; the initial associated information is unprocessed information associated with the processing process of the target task; the initial associated information is stored in a first sentence granularity; the first sentence granularity represents the length of information;

[0007] The initial association information is split to generate a plurality of first association information of a second sentence granularity; the second sentence granularity represents the length of the information, and the second sentence granularity is smaller than the first sentence granularity; the first association information is information with complete semantics;

[0008] Extracting features from the target problem and each of the first associated information, and determining the degree of association between the extracted features as the degree of association between the target task and each of the first associated information; the degree of association is used to characterize the degree of similarity between the two pieces of information;

[0009] Selecting the first association information whose association degree is greater than a first threshold from each of the first association information as the second association information;

[0010] An information extraction process of content summary is performed on each of the second related information to generate target related information corresponding to the target task.

[0011] In a second aspect, an embodiment of the present disclosure further provides an information processing device, the device comprising:

[0012] The target task acquisition module is used to acquire the target task to be processed and initial associated information; the initial associated information is unprocessed information that is associated with the processing process of the target task; the initial associated information is stored in units of a first sentence granularity; the first sentence granularity represents the length of information;

[0013] A first association information generating module is used to split the initial association information to generate a plurality of first association information of a second sentence granularity; the second sentence granularity represents the length of information, and the second sentence granularity is smaller than the first sentence granularity; the first association information is information with complete semantics;

[0014] a correlation determination module, used to extract features from the target problem and each of the first correlation information, and determine the correlation between the extracted features as the correlation between the target task and each of the first correlation information; the correlation is used to represent the similarity between the two pieces of information;

[0015] A second association information screening module, configured to screen the first association information whose association degree is greater than a first threshold from each of the first association information as the second association information;

[0016] The target-related information generating module is used to extract and process the content of each of the second-related information to generate target-related information corresponding to the target task.

[0017] In a third aspect, an embodiment of the present disclosure further provides an electronic device, the electronic device comprising:

[0018] Processor and memory;

[0019] The processor is used to execute the information processing method described in any embodiment of the present disclosure by calling the program or instruction stored in the memory.

[0020] In a fourth aspect, an embodiment of the present disclosure further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a program or instructions, wherein the program or the instructions enable a computer to execute the information processing method described in any embodiment of the present disclosure.

[0021] The information processing scheme provided by the disclosed embodiment, after acquiring the target task and the initial associated information stored in the first sentence granularity, splits the initial associated information to generate multiple first associated information of the second sentence granularity with a smaller sentence granularity; it realizes the disassembly of the initial associated information with a longer information length into each first associated information with a shorter information length, shortens the length of the associated information participating in the target task matching each time, so that the matching and screening of the associated information can be performed more finely, and the accuracy of obtaining the associated information is improved. By determining the correlation between the target task and each first associated information, and screening the second associated information with a correlation greater than the first threshold from each first associated information; it realizes the screening of some first associated information with a higher correlation with the target task from a large number of first associated information, and eliminates the associated information with a relatively low correlation, thereby reducing the amount of data in the information processing process, and further reducing the computational cost and communication overhead of the subsequent processing of the target associated information. The target-related information corresponding to the target task is generated through information extraction processing based on the content summary of each second-related information; the processing of summarizing and generalizing part of the screened first-related information is realized to avoid irrelevant information occupying information space, which can further reduce the data volume of the target-related information, and at the same time, more key information related to the target task is retained in the target-related information, reducing the probability of losing key information, thereby further improving the accuracy of the target-related information and reducing the impact of the related information on the accuracy and efficiency of the system response. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0023] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0024] Figure 1 A flowchart of an information processing method provided by an embodiment of the present disclosure;

[0025] Figure 2 A data flow diagram of an information processing method provided by an embodiment of the present disclosure;

[0026] Figure 3 A flowchart of another information processing method provided by an embodiment of the present disclosure;

[0027] Figure 4 A schematic diagram of the structure of an information processing device provided by an embodiment of the present disclosure;

[0028] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0029] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described in detail below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.

[0030] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0031] The information processing method provided by the embodiments of the present disclosure is mainly applicable to scenarios with large amounts of information interaction. Exemplarily, the information processing method can be applicable to scenarios in which related information is obtained during task processing in a multi-agent system (MAS) or a large language model, and can also be applicable to scenarios in which a distributed system obtains related data from a database to participate in big data computing. For example, it can be specifically applied to application scenarios such as intelligent customer service, collaborative robots, and multimodal dialogue systems. The above-mentioned multi-agent system is a distributed artificial intelligence system composed of multiple independent intelligent agents, which work together through these intelligent agents to share and process complex tasks. The above-mentioned agent refers to an entity that can act autonomously, perceive the environment, and make decisions, such as a software program (such as a large-scale machine learning model such as GPT), a robot, or other forms of intelligent systems.

[0032] The information processing method provided by the embodiments of the present disclosure may be executed by an information processing device, which may be implemented by software and / or hardware, and which may be integrated into an electronic device with certain data processing capabilities, such as but not limited to a smart phone, a laptop computer, a desktop computer or a server.

[0033] Figure 1 is a flow chart of an information processing method provided by an embodiment of the present disclosure. Figure 1 , the information processing method specifically includes:

[0034] S110: Obtain the target task to be processed and initial associated information.

[0035] Among them, the target task is the task to be processed by obtaining related information. The target task can be a general task input by the user or issued by the upstream business, or it can be a subtask with smaller granularity and more single content semantics obtained by decomposing the general task. The task in the embodiment of the present disclosure can be a question / question that needs to be answered or explained, or it can be a description of an imperative task that needs to be completed (such as writing an article, creating multimedia content, calculating a certain indicator, etc.).

[0036] Among them, the initial associated information is all the complete associated information contained in the database. The associated information in the embodiment of the present disclosure is unprocessed information that has an associated relationship with the processing process of the target task. It is used to complete the information required for processing the target task and is existing information that has a certain associated relationship with the target task. The associated information can be a corpus with semantics / contextual relevance in the field of natural language, for example, it can be the text content of the previous input that is logically related to the question currently being asked; it can also be data with an associated relationship in the process of big data processing, for example, it can be those states and parameters that are transmitted between nodes in a distributed system and are related to completing a certain function or processing a certain request. The initial associated information is stored in units of the first sentence granularity. The first sentence granularity represents the length of the information. For example, for the associated information of the sentence class, the first sentence granularity can be a relatively long sentence length such as a chapter or paragraph.

[0037] Specifically, after starting the function of acquiring associated information (such as a large language model starting the process of answering questions, or a big data system starting to process a data processing task, etc.), the electronic device can obtain the target task from the human-computer interaction interface or upstream business. Alternatively, the electronic device obtains the total task in the above manner, and then decomposes it to obtain the target task. Alternatively, after other functional modules execute the decomposition logic of the total task, the electronic device obtains the target task after the decomposition of the total task from the functional module.

[0038] In addition, the electronic device may obtain the initial association information from a storage space storing association information (such as a local storage space or a cloud storage space on a network end, etc.).

[0039] S120: Split the initial association information to generate a plurality of first association information of a second sentence granularity.

[0040] Among them, the second sentence granularity is a pre-set length used to characterize information. The second sentence granularity is smaller than the first sentence granularity. In one example, the second sentence granularity can be a relatively long sentence length, such as a paragraph with relatively independent semantics that is shorter than the chapter or paragraph of the first sentence granularity. In another example, the second sentence granularity can be a relatively short sentence length, such as a partial sentence, a sentence, or a few sentences with relatively complete semantics. The first associated information is the local content in the initial associated information, which is information with complete semantics.

[0041] Specifically, the initial associated information is complete associated information, which contains more information. If the target task is used to directly search for the initial associated information, there will be a long and large amount of matching content, resulting in a lengthy process of obtaining associated information, a large loss of computing power, and low efficiency in information acquisition. Based on this, the initial associated information is split in the embodiment of the present disclosure, and the lengthy initial associated information is broken down into multiple first associated information with a smaller second sentence granularity. In this way, the electronic device can match the target task with each first associated information with a shorter content to determine whether the first associated information is valid information related to the target task, and achieve more refined screening and matching of associated information, thereby improving the accuracy and efficiency of information acquisition.

[0042] In some embodiments, S120 includes: performing semantic-based sentence processing on the initial association information to generate multiple fifth association information of third sentence granularity; performing grammatical-based sentence processing on any fifth association information to generate multiple first association information of second sentence granularity.

[0043] Among them, the third sentence granularity is another pre-set sentence length, which is used to characterize the length of information. In this embodiment, the second sentence granularity is a relatively short sentence length, and the third sentence granularity is a relatively long sentence length, that is, the third sentence granularity is smaller than the first sentence granularity, and the third sentence granularity is larger than the second sentence granularity. For example, the first sentence granularity corresponds to the length of a chapter, the second sentence granularity corresponds to the length of a small number of sentences, and the third sentence granularity corresponds to the length of a paragraph containing multiple sentences. The fifth associated information is information with complete semantics.

[0044] Specifically, the process of disassembling the initial associated information by the electronic device may be: firstly, the initial associated information is processed based on semantics, and the associated information (i.e., the fifth associated information) with relatively weak semantic relevance between each other and large semantic relevance of internal words and the third sentence granularity is generated. In one example, the electronic device can use the sentence algorithm in natural language processing (NLP), such as the maximum matching method, machine learning model, etc., to process the initial associated information by sentence, and obtain multiple fifth associated information to improve the accuracy of information segmentation. In another example, the electronic device can use the processing method of word segmentation and sentence segmentation to cut the complex initial associated information into several semantically related first associated information of the third sentence granularity. For example, the initial associated information can be processed by a word segmentation tool such as jieba word segmentation to obtain multiple words. Then, the relevant algorithms of syntax and semantic analysis are used to semantically analyze, combine and organize these words to generate multiple fifth associated information, so as to reduce the storage space and computing resources required by the machine learning model on the basis of ensuring the accuracy of information segmentation, thereby reducing the cost of sentence segmentation processing.

[0045] Then, the electronic device continues to perform sentence processing on the relatively long fifth associated information to obtain the first associated information with a smaller second sentence granularity. The electronic device may consider the grammar of the sentence and use a text segmentation algorithm to further refine the fifth associated information. For example, the natural boundaries of the associated information (such as punctuation marks, keywords, etc.) may be combined to perform sentence processing using regular expressions or rule-based methods. This can further shorten the sentence length of the associated information involved in the matching, thereby further improving the sophistication of the associated information screening and matching, and further improving the accuracy and efficiency of the associated information acquisition.

[0046] See also Figure 2 In the data flow illustrated by the solid line, after the electronic device performs sentence processing on the initial associated information, refined first associated information 1-1, first associated information 1-2, ..., first associated information n-1, ..., first associated information nz can be obtained. The first dimension 1 to n represents the sequence number of the fifth associated information, and the second dimension 1 to x / y / z represents the sequence number of the first associated information after a certain fifth associated information is disassembled.

[0047] In some other embodiments, S120 can also be implemented as: performing sentence processing based on grammar on the initial association information to generate multiple fifth association information of the third sentence granularity; performing sentence processing based on semantics on any fifth association information to generate multiple first association information of the second sentence granularity. That is, the electronic device can set the sentence processing in the above process to a process based on grammar processing first and then semantic processing. In this way, the complete initial association information can be split into sentences first, and then these sentences can be integrated and processed according to semantics to obtain the final multiple first association information.

[0048] In some other embodiments, S120 may also be implemented as: inputting the initial associated information into a pre-trained information splitting model, and splitting the initial associated information into a plurality of first associated information after being processed by the model. The information splitting model is a machine learning model (such as a large language model) having the function of splitting information according to the second sentence granularity.

[0049] S130 , extracting features from the target task and each piece of first association information, and determining the degree of association between the extracted features as the degree of association between the target task and each piece of first association information.

[0050] The correlation degree is used to characterize the similarity between two pieces of information. In the disclosed embodiment, the correlation degree indicates the correlation degree between the target task and the first associated information. The greater the correlation degree, the higher the correlation degree, and the more likely the corresponding first associated information is the information required to process the target task.

[0051] Specifically, in order to reduce the amount of data, the electronic device can filter out the first associated information that has a certain degree of relevance to the target task from each first associated information, and eliminate some redundant associated information with a lower degree of relevance. Therefore, the electronic device can calculate the degree of association between the target task and each first associated information. For example, the electronic device can use a deep learning model such as a bidirectional encoder representation from transformers (BERT) or other transformer-based deep learning models to extract features and calculate feature similarity of the input target task and the first associated information to obtain the degree of association corresponding to the first associated information.

[0052] See also Figure 2 In the data flow illustrated by the solid line, the electronic device can calculate the degree of association between the target task 1 and each first association information. If you want to obtain the association information adapted to each target task, or obtain the association information corresponding to the total task / initial task, then you can obtain the degree of association between each target task i (i ranges from 1 to m) and each first association information according to the above process.

[0053] S140: Filter out first association information whose association degree is greater than a first threshold from each piece of first association information as second association information.

[0054] The first threshold is a preset relevance threshold, which is used to filter the first relevance information related to the target task.

[0055] Specifically, the electronic device can compare the degree of association of each first associated information with the first threshold value to filter out the first associated information with a lower degree of association, while retaining the first associated information with a higher degree of association. If the comparison result is that the degree of association is less than or equal to the first threshold value, the first associated information corresponding to the degree of association is not selected. If the comparison result is that the degree of association is greater than the first threshold value, it means that the first associated information corresponding to the degree of association is an important information fragment related to the target task, and then the first associated information is screened out as the second associated information. At this point, a batch of second associated information that is more important to the target task is screened out from all the first associated information corresponding to the initial associated information.

[0056] S150 , performing information extraction processing for content summary on each piece of second related information to generate target related information corresponding to the target task.

[0057] The target-related information is the related information that ultimately participates in the processing of the target task.

[0058] Specifically, the second association information is still information of a certain length, in which there may be redundant information, and there may be repeated information between the second association information. Therefore, in order to further reduce unnecessary token usage, thereby further reducing system overhead while retaining key information, see Figure 2 In the data flow illustrated by the solid line, in the embodiment of the present disclosure, the screened second associated information can be summarized and generalized for information extraction processing to obtain key information in the initial associated information that is highly relevant to the target task as the target associated information.

[0059] The information extraction and processing method of the above-mentioned content summary can be to directly input each second related information into a machine learning model with content summarization function to obtain the target related information; it can also be to first extract key information from each second related information, and then reorganize the extracted key information to obtain the target related information; it can also be to first extract features from each second related information, and then input the extracted features into the model, and finally reorganize the results output by the model to obtain the target related information.

[0060] The information processing method provided by the above-mentioned embodiments of the present disclosure, after acquiring the target task and the initial associated information stored in the second sentence granularity, splits the initial associated information to generate multiple first associated information of the third sentence granularity with a smaller sentence granularity; realizes the disassembly of the initial associated information with a longer information length into each first associated information with a shorter information length, shortens the length of the associated information participating in the target task matching each time, so that the matching and screening of the associated information can be performed more finely, and the accuracy of obtaining the associated information is improved. By determining the correlation between the target task and each first associated information, and screening the second associated information with a correlation greater than the first threshold from each first associated information; it is realized that part of the first associated information with a higher correlation with the target task is screened out from a large number of first associated information, and the associated information with a relatively low correlation is eliminated, thereby reducing the amount of data in the information processing process, and then reducing the computational cost and communication overhead of the subsequent processing of the target associated information. The target-related information corresponding to the target task is generated through information extraction processing based on the content summary of each second-related information; the processing of summarizing and generalizing part of the filtered first-related information is realized to avoid irrelevant information occupying information space, which can further reduce the data volume of the target-related information, and at the same time, the target-related information retains more key information related to the target task, reducing the probability of losing key information, thereby further improving the accuracy of the target-related information and reducing the impact of the related information on the accuracy and efficiency of the system response. Second statement granularity.

[0061] In some embodiments, Figure 1 After S140 of the information processing method shown, the information processing method further includes: selecting first related information having a degree of association greater than a second threshold from each piece of first related information as third related information.

[0062] The second threshold is another preset relevance threshold, which is used to filter the first relevance information that is more relevant to the target task. The second threshold is greater than the first threshold.

[0063] Specifically, the key information obtained by extracting the information from each second associated information can largely represent the information in the initial associated information that has a high relevance to the target task, but due to its summary processing, there may still be a certain degree of information loss. In order to further reduce the information loss without increasing the amount of data too much, a second threshold with a larger value can be set in this embodiment to filter out the first associated information with a higher degree of relevance to the target task from each first associated information, and to complete the summarized key information.

[0064] In a specific implementation, the electronic device may compare the degree of association of each first associated information with the second threshold. If the comparison result is that the degree of association is less than or equal to the second threshold, the first associated information corresponding to the degree of association is not selected. If the comparison result is that the degree of association is greater than the second threshold, it means that the first associated information corresponding to the degree of association is a more important information fragment with a great similarity to the target task, then the first associated information is screened out as the third associated information. At this point, at least one third associated information that is more important to the target task is screened out from all the first associated information corresponding to the initial associated information.

[0065] See also Figure 2 In the data flow example connected by dotted lines, the electronic device compares the degree of association between the target task 1 and each first associated information according to the second threshold value to filter out the third associated information.

[0066] In some embodiments, based on the above embodiments, Figure 1 S150 of the information processing method shown includes: performing information extraction processing for content summarization on each second related information to generate fourth related information; and determining a union of the third related information and the fourth related information as target related information corresponding to the target task.

[0067] Specifically, see Figure 2 In the example data flow connected by dotted lines, the electronic device performs information extraction processing to summarize the screened second associated information, and obtains the key information in the initial associated information that is highly relevant to the target task, which is called the fourth associated information. Then, the residual idea is introduced, and the electronic device fuses the third associated information obtained by the above screening and the fourth associated information to obtain the target associated information. In this way, the extracted key information and the screened associated information that is more relevant to the target task 1 can be retained at the same time, further reducing the risk of information loss and enhancing the integrity and accuracy of the target associated information.

[0068] In this embodiment, the third associated information and the fourth associated information may be merged by directly concatenating the two information, that is, obtaining the union of the third associated information and the fourth associated information, so that the target associated information completely retains the two information, which can improve the information content of the target associated information.

[0069] In other embodiments, based on the above embodiments, Figure 1S150 of the information processing method shown includes: performing information extraction processing for content summary of each second related information to generate fourth related information; comparing the difference in information volume of the third related information and the fourth related information as the information difference between the third related information and the fourth related information; determining a weight value based on the information difference, and performing weighted fusion processing on the third related information and the fourth related information based on the weight value to generate target related information corresponding to the target task.

[0070] Specifically, the fusion method of the third associated information and the fourth associated information can be to perform weighted fusion of the two information, and the weighted weight is determined based on the degree of difference between the two information (i.e., the information difference). In this embodiment, the amount of information of the two information can be compared, and then the weighted weight can be set proportionally according to the value of the information difference. For example, if the amount of information of the third associated information is more, then a larger weighted weight can be set for the third associated information according to the proportion of the information difference, and a relatively smaller weighted weight can be set for the fourth associated information. Then, the amount of information of the third associated information and the fourth associated information are weighted and summed according to these weighted weights to obtain the target associated information. In this way, more associated information with high information content can be retained in the target associated information, and the amount of data of the target associated information will not be increased much.

[0071] The amount of information can be characterized by at least one of the length of the information, the size of the storage space occupied by the information, the complexity of the information structure, and the redundancy of the information. The complexity of the information structure is characterized by at least one of the number of elements contained in the information, the type of elements, the hierarchical relationship between the elements, and the logical association between the elements. The redundancy of the information is characterized by the proportion of repeated information (such as repeated words or repeated meanings) and / or the proportion of key information.

[0072] In some further embodiments, based on the above embodiments, Figure 1 S150 of the information processing method shown includes: performing information extraction processing for content summarization on each second related information to generate fourth related information; performing keyword matching on the third related information and the fourth related information, and determining the information difference between the third related information and the fourth related information based on the keywords that did not match successfully; determining a weight value based on the information difference, and performing weighted fusion processing on the third related information and the fourth related information based on the weight value to generate target related information corresponding to the target task.

[0073] Specifically, the fusion method of the third associated information and the fourth associated information can be to perform weighted fusion of the two information, and the weighted weight is determined based on the information difference of the two information. In this embodiment, the two information can be compared from the perspective of information content to determine the information difference of the two information. That is, the third associated information and the fourth associated information are keyword matched to obtain successfully matched keywords and unmatched keywords. Then, the number or proportion of unmatched keywords is calculated as the information difference of the two information. After that, the weighted weight is set in proportion according to the value of the information difference. For example, if there are many unmatched keywords or their proportion exceeds a certain value, it means that the difference between the two information is large, and more key information may be omitted in the fourth associated information. In this case, a larger weighted weight can be set for the third associated information, and a relatively smaller weighted weight can be set for the fourth associated information, and then the two information are fused according to these weighted weights to obtain the target associated information. In this way, the content of key information in the target associated information can be further improved.

[0074] In some further embodiments, based on the above embodiments, Figure 1 S150 of the information processing method shown includes: performing information extraction processing for content summarization on each second related information to generate fourth related information; comparing the difference in information volume between the third related information and the fourth related information, and performing keyword matching on the third related information and the fourth related information, and determining the information difference between the third related information and the fourth related information based on the difference in information volume and the keywords that were not successfully matched; determining a weight value based on the information difference, and performing weighted fusion processing on the third related information and the fourth related information based on the weight value to generate target related information corresponding to the target task.

[0075] Specifically, the fusion method of the third associated information and the fourth associated information can be to perform weighted fusion of the two information, and the weighted weight is determined based on the information difference between the two information. In this embodiment, the two information can be compared from the two perspectives of information volume and information content according to the description of the above embodiments to determine the information difference between the two information. Afterwards, the weighted weight is set proportionally according to the value of the information difference. For example, if the amount of information of the third associated information is large, and there are many keywords that have not been successfully matched or their proportion exceeds a certain value, a larger weighted weight can be set for the third associated information, and a relatively smaller weighted weight can be set for the fourth associated information, and then the two information are fused according to these weighted weights to obtain the target associated information. In this way, the key information can be better and more completely retained, and the content of effective information in the target associated information can be further improved.

[0076] In some embodiments, the above-mentioned information extraction processing of summarizing the content of each second related information includes: taking the relevance corresponding to the second related information as the attention weight, and utilizing an information extraction algorithm based on the attention mechanism to perform information extraction processing on the content of each second related information.

[0077] Specifically, see Figure 2 In the data flow illustrated by the dotted dashed line, when the electronic device extracts information summarizing and generalizing each second related information, it can adopt an information extraction method based on the attention mechanism, in which the attention weight is the correlation corresponding to the second related information. In this way, when extracting information, more information will be extracted from the second related information with a higher correlation, and relatively less information will be extracted from the second related information with a relatively lower correlation, so that the finally summarized and generalized related information contains more and more important information, further improving the accuracy of the final target related information.

[0078] In some embodiments, Figure 1 In S110 of the information processing method shown, "obtaining a target task to be processed" includes: obtaining an initial task to be processed, and determining the initial task as the target task. In this embodiment, the obtained total task (ie, the initial task) is used as the target task to obtain associated information.

[0079] In other embodiments, Figure 1 In S110 of the illustrated information processing method, “obtaining a target task to be processed” includes: obtaining an initial task to be processed; splitting the initial task to generate a plurality of subtasks corresponding to the initial task; and determining any subtask as a target task.

[0080] Specifically, the electronic device receives an initial task input by a user or issued by an upstream business. Then, the electronic device splits the initial task and cuts the complex initial task into several subtasks with simpler task contents. The task content of each subtask has complete semantics. The process of the splitting process can be found in the description of the relevant embodiments of the fifth association information and the first association information in S120 above.

[0081] See also Figure 2 In the data flow illustrated by the dashed line, after the electronic device splits the initial task, m subtasks can be obtained, and each subtask can be used as a target task to obtain the associated information of the above embodiments, so that target task 1, target task 2, ..., target task m can be obtained.

[0082] In some embodiments, based on the above embodiment of decomposing the initial task into multiple target tasks, Figure 1After S150 of the information processing method shown, the information processing method further includes: aggregating target association information corresponding to each target task to generate summary association information corresponding to the initial task.

[0083] Specifically, the electronic device can execute the processing of obtaining associated information in the above embodiments for each target task, and obtain the target associated information corresponding to each target task. If the business requirement is to obtain the associated information of the initial task, then the electronic device can aggregate the obtained target associated information to obtain comprehensive target associated information adapted to the initial task, that is, summarized associated information. The summarized associated information can participate in the processing of the initial task. For example, when the agent processing the initial task and the associated information are not on the same device, the summarized associated information can be transmitted to the agent; when the agent and the associated information are on the same device, the agent can directly obtain the locally obtained summarized associated information.

[0084] See also Figure 2 In the data flow illustrated by the dashed line, for the target task m, the electronic device can also calculate the degree of association between it and each first associated information. Then, at least one second associated information is screened out by the first threshold, and at least one third associated information is screened out by the second threshold. Then, the degree of association corresponding to the second associated information is used as the attention weight, and the information extraction process based on the content summarization of the attention mechanism is performed on each second associated information to obtain the fourth associated information adapted to the target task m. Finally, the third associated information and the fourth associated information corresponding to the target task m are added to the fusion process.

[0085] According to the above process, the electronic device can obtain the third association information and the fourth association information adapted to each target task ( Figure 2 The processing flow for other target tasks is not shown. Then, all the third association information and the fourth association information are fused to obtain the summary association information adapted to the initial task.

[0086] Figure 3 is a flowchart of another information processing method provided by an embodiment of the present disclosure. The explanations of the terms that are the same or corresponding to the above embodiments are not repeated here. Figure 3 , the information processing method comprises:

[0087] S310: Obtain the target task to be processed and initial associated information.

[0088] S320: Analyze the target task from at least one task dimension, determine the analysis result of the task dimension, and determine the complexity of the target task based on the analysis result of the task dimension.

[0089] The task dimension is a dimension that reflects the complexity of the task. For example, the task dimension includes the amount of task data, the amount of parameters involved in the task, the number of constraints involved in the task, the number of subtasks obtained by decomposing the task, or the dependencies between subtasks.

[0090] Specifically, different target tasks require different associated information, and the corresponding computational cost and communication cost of the associated information are also different. Therefore, before obtaining the target associated information, the electronic device can first analyze the target task and its corresponding initial associated information to determine the appropriate way to process the initial associated information according to the analysis result.

[0091] During specific implementation, the electronic device can analyze the target task from at least one task dimension to obtain the analysis result of the corresponding task dimension. For example, the electronic device can analyze the amount of data of the target task, the amount of parameters involved in the target task, whether the target task involves constraints and the number of constraints, the number of subtasks obtained after statistically decomposing the target task, the degree of dependence between each subtask or the depth of the dependency relationship, etc., and obtain the corresponding analysis result. Then, if there is only one analysis result of a task dimension, the electronic device can determine it as the complexity of the target task. If there are analysis results of multiple task dimensions, the electronic device can normalize these analysis results and weight each normalized result (the weight can be set according to business needs), and the weighted result obtained is the complexity of the target task.

[0092] S330: Analyze the initial associated information from at least one information volume dimension, determine the analysis result of the information volume dimension, and determine the information volume of the initial associated information based on the analysis result of the information volume dimension.

[0093] Among them, the information volume dimension is a dimension that reflects the amount of information. Exemplarily, the information volume dimension includes the length of information, the size of the storage space occupied by the information, the complexity of the information structure, or the information redundancy. The complexity of the information structure is characterized by at least one of the number of elements contained in the information, the type of elements, the hierarchical relationship between the elements, and the logical association between the elements. The information redundancy is characterized by the proportion of repeated information and / or the proportion of key information.

[0094] Specifically, the electronic device can count the information length of the initial associated information, count the size of the storage space occupied by the initial associated information, analyze the structural complexity of the initial associated information, and count the information redundancy of the initial associated information to obtain the corresponding analysis result. Then, if there is only one analysis result of the information volume dimension, the electronic device can determine it as the information volume of the initial associated information. If there are multiple analysis results of the information volume dimension, the electronic device can normalize these analysis results and weight each normalized result (the weight can be set according to business requirements), and the obtained weighted result is the information volume of the initial associated information.

[0095] S340, based on the complexity of the target task and / or the amount of information in the initial associated information, determine an information processing method; the information processing method includes retaining the initial associated information, extracting information summarizing the content of the initial associated information, or triggering execution Figure 1 S120 is shown.

[0096] The information processing method is a method for processing the initial associated information.

[0097] Specifically, the electronic device may determine the information processing method for processing the initial associated information according to the complexity of the target task and / or the amount of information in the initial associated information. For example, the more complex the target task is and the larger the amount of information in the initial associated information is, the more complex the information processing method is, so as to ensure that the target associated information has an appropriate amount of data and more key information.

[0098] In some embodiments, the process of determining the information processing method is: if the amount of information of the initial association information is less than a third threshold, determining the information processing method is to retain the initial association information.

[0099] The third threshold is a preset information volume threshold, and its value is adapted to the information volume dimension. For example, when the information volume dimension is information length represented by sentence length, the third threshold can be a predetermined total sentence length of the associated information.

[0100] Specifically, the electronic device compares the amount of information of the initial association information with the third threshold. If the amount of information is greater than or equal to the third threshold, it means that the amount of information of the initial association information is large, and directly using the initial association information as the target association information will cause a sharp increase in token usage, which in turn leads to a significant increase in computing costs and communication overheads. In this case, other information processing methods can be used. If the above-mentioned amount of information is less than the third threshold, it means that the amount of information of the initial association information is small, and directly using the initial association information as the target association information will not cause a significant change in token usage, computing costs, and communication overheads. Therefore, the information processing method in this case is to retain the initial association information.

[0101] In other embodiments, the process of determining the information processing method is: if the amount of information of the initial associated information is greater than or equal to the third threshold, and the complexity of the target task is less than the fourth threshold, then the information processing method is determined to be information extraction processing for content summarization of the initial associated information.

[0102] Among them, the fourth threshold is a pre-set complexity threshold, and its value is adapted to the task dimension.

[0103] Specifically, if the amount of information in the initial associated information is greater than or equal to the third threshold, the initial associated information cannot be used directly. In this case, the complexity of the target task is further evaluated. If the complexity of the target task is less than the fourth threshold, it means that the target task is relatively simple and the electronic device can easily and accurately understand the target task. Then, it is relatively easy and accurate to extract the associated information that is more relevant to the target task from the initial associated information. Therefore, the electronic device can determine the information processing method as information extraction processing of the initial associated information.

[0104] In some other embodiments, the process of determining the information processing method is: if the amount of information in the initial associated information is greater than or equal to a third threshold, and the complexity of the target task is greater than or equal to a fourth threshold, then determining the information processing method is to trigger execution Figure 1 The subsequent flow of the information processing method is shown.

[0105] Specifically, according to the above description, if the amount of information in the initial associated information is greater than or equal to the third threshold, and the complexity of the target task to be further evaluated is greater than or equal to the fourth threshold, it means that the target task is relatively complex and a comprehensive summary of the initial associated information is likely to lose important information. In this case, the electronic device determines the information processing method as Figure 1 And the information processing methods shown in the above embodiments.

[0106] S350: Process the initial associated information according to the information processing method to obtain target associated information.

[0107] Specifically, the electronic device may process the initial association information according to a determined information processing method to obtain target association information adapted to the target task.

[0108] In some embodiments, S350 is implemented as S351: the information processing method is to retain the initial association information, and the electronic device uses the initial association information as the target association information.

[0109] In some other embodiments, S350 is implemented as S352: the information processing method is to perform information extraction processing for summarizing the content of the initial associated information, and the electronic device may perform information extraction processing for summarizing the initial associated information.

[0110] In one example, the electronic device may input the initial association information into a relevant information extraction model to obtain the target association information. In another example, in order to improve the accuracy of the information, the electronic device may consider its relevance to the target task during the process of extracting information from the initial association information. For example, the above embodiments may be referred to to calculate the correlation between the target task and the local information in the initial association information, and then the correlation may be used as the attention weight to perform information extraction processing based on the attention mechanism to obtain the target association information. For another example, the target task and the initial association information may be directly input into an information extraction model based on the attention mechanism to obtain the target association information.

[0111] In some other embodiments, S350 is implemented as follows: the information processing method is to trigger execution Figure 1 In the steps S120 to S150 shown, the electronic device may execute Figure 1 And the information processing methods shown in the above embodiments.

[0112] The information processing method provided by the above-mentioned embodiments of the present disclosure determines, based on the complexity of the target task and / or the amount of information of the initial associated information, an information processing method as retaining the initial associated information, performing information extraction processing on the initial associated information, or triggering the execution of splitting processing on the initial associated information to generate a plurality of first associated information of the second sentence granularity; then, the initial associated information is processed according to the information processing method to obtain the target associated information; and dynamically determines the processing method of the associated information adapted to the target task according to preset rules and real-time monitoring data of the target task and the initial associated information, so that the information processing method has greater flexibility to adapt to the needs of different application scenarios, thereby better balancing the information integrity, accuracy and acquisition efficiency of the associated information, further saving resources, greatly improving the response speed and processing flexibility of the system, and thereby improving the adaptability of the system to the needs of different application scenarios.

[0113] Figure 4 The structure diagram of an information processing device provided by the embodiment of the present disclosure is shown in FIG. Figure 4 As shown, the information processing device 400 includes:

[0114] The target task acquisition module 410 is used to acquire the target task to be processed and initial associated information; the initial associated information is the unprocessed information associated with the processing process of the target task; the initial associated information is stored in the first sentence granularity; the first sentence granularity represents the length of the information;

[0115] The first associated information generating module 420 is used to split the initial associated information to generate a plurality of first associated information of a second sentence granularity; the second sentence granularity represents the length of the information, and the second sentence granularity is smaller than the first sentence granularity; the first associated information is information with complete semantics;

[0116] The correlation determination module 430 is used to extract features from the target problem and each first correlation information, and determine the correlation between the extracted features as the correlation between the target task and each first correlation information; the correlation is used to represent the similarity between two pieces of information;

[0117] The second association information screening module 440 is used to screen out first association information with a degree of association greater than a first threshold from each piece of first association information as second association information;

[0118] The target-related information generating module 450 is used to perform information extraction processing for content summarization on each piece of second related information to generate target-related information corresponding to the target task.

[0119] An information processing device provided by the embodiment of the present disclosure, after acquiring the target task and the initial associated information stored in the second sentence granularity, splits the initial associated information to generate multiple first associated information of the third sentence granularity with a smaller sentence granularity; the initial associated information with a longer information length is split into each first associated information with a shorter information length, shortening the length of the associated information participating in the target task matching each time, so that the associated information can be matched and screened more finely, and the accuracy of acquiring the associated information is improved. By determining the correlation between the target task and each first associated information, and screening the second associated information with a correlation greater than the first threshold from each first associated information; it is achieved to screen out some first associated information with a higher correlation with the target task from a large number of first associated information, and eliminate the associated information with a relatively low correlation, thereby reducing the amount of data in the information processing process, and then reducing the computational cost and communication overhead of subsequent processing of the target associated information. The target-related information corresponding to the target task is generated through information extraction processing based on the content summary of each second-related information; the processing of summarizing and generalizing part of the filtered first-related information is realized to avoid irrelevant information occupying information space, which can further reduce the data volume of the target-related information, and at the same time, the target-related information retains more key information related to the target task, reducing the probability of losing key information, thereby further improving the accuracy of the target-related information and reducing the impact of the related information on the accuracy and efficiency of the system response. Second statement granularity.

[0120] In some embodiments, the information processing device 400 further includes a third associated information screening module, which is used to:

[0121] After selecting second association information whose association degree is greater than a first threshold from each first association information, selecting first association information whose association degree is greater than a second threshold from each first association information as third association information; the second threshold is greater than the first threshold;

[0122] Accordingly, the target association information generating module 450 is specifically used for:

[0123] Performing information extraction processing on content summary of each second related information to generate fourth related information;

[0124] Determine a union of the third association information and the fourth association information as target association information corresponding to the target task;

[0125] Alternatively, comparing the difference between the amount of information of the third related information and the amount of information of the fourth related information, as the information difference between the third related information and the fourth related information; and / or, performing keyword matching on the third related information and the fourth related information, and determining the information difference between the third related information and the fourth related information based on keywords that have not been successfully matched;

[0126] A weight value is determined based on the information difference, and a weighted fusion process is performed on the third association information and the fourth association information based on the weight value to generate target association information corresponding to the target task.

[0127] In some embodiments, the target association information generating module 450 is further specifically used to:

[0128] The relevance degree corresponding to the second related information is used as the attention weight, and an information extraction algorithm based on the attention mechanism is used to perform information extraction processing on the content summary of each second related information.

[0129] In some embodiments, the first association information generating module 420 is specifically used to:

[0130] The initial associated information is processed based on semantics to generate a plurality of fifth associated information of the third sentence granularity; the third sentence granularity represents the length of the information; the third sentence granularity is smaller than the first sentence granularity, and the third sentence granularity is larger than the second sentence granularity; the fifth associated information is information with complete semantics;

[0131] A grammar-based sentence segmentation process is performed on any fifth association information to generate a plurality of first association information of a second sentence granularity.

[0132] In some embodiments, the information processing device 400 further includes an information processing mode determination module, which is used to:

[0133] After obtaining the target task to be processed and the initial associated information, analyzing the target task from at least one task dimension, determining the analysis result of the task dimension, and determining the complexity of the target task based on the analysis result of the task dimension; wherein the task dimension is a dimension reflecting the complexity of the task;

[0134] Analyze the initial associated information from at least one information volume dimension, determine the analysis result of the information volume dimension, and determine the information volume of the initial associated information based on the analysis result of the information volume dimension; wherein the information volume dimension is a dimension reflecting the amount of information;

[0135] Based on the complexity of the target task and / or the amount of information in the initial associated information, determine an information processing method for the initial associated information; wherein the information processing method includes retaining the initial associated information, performing information extraction processing on the initial associated information to summarize the content, or triggering the execution of sentence processing on the initial associated information to generate a plurality of first associated information of a second sentence granularity;

[0136] The initial associated information is processed according to the information processing method to obtain target associated information.

[0137] In some embodiments, the information processing mode determination module is specifically used to:

[0138] If the amount of information of the initial association information is less than the third threshold, determining the information processing method is to retain the initial association information;

[0139] If the amount of information of the initial associated information is greater than or equal to the third threshold, and the complexity of the target task is less than the fourth threshold, determining the information processing method to be information extraction processing for content summarization of the initial associated information;

[0140] If the amount of information in the initial associated information is greater than or equal to the third threshold, and the complexity of the target task is greater than or equal to the fourth threshold, the information processing method is determined to trigger the execution of splitting the initial associated information to generate multiple first associated information of the second statement granularity.

[0141] In some embodiments, the target task acquisition module 410 is specifically used to:

[0142] Obtaining the initial task to be processed, and determining the initial task as the target task;

[0143] In some other embodiments, the target task acquisition module 410 is specifically used to:

[0144] Get the initial tasks to be processed;

[0145] The initial task is split to generate multiple subtasks corresponding to the initial task; wherein the task contents of the subtasks have complete semantics;

[0146] Identify any subtask as the target task.

[0147] In some embodiments, the information processing device 400 further includes a summary association information generation module, which is used to:

[0148] After performing information extraction processing for content summarization on each second related information to generate target related information corresponding to the target task, the target related information corresponding to each target task is aggregated to generate summary related information corresponding to the initial task.

[0149] The information processing device provided by the embodiments of the present disclosure can execute the information processing method provided by any embodiment of the present disclosure, and has the corresponding functional modules and beneficial effects of the execution method.

[0150] It is worth noting that in the embodiment of the above-mentioned information processing device, the various modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be realized; in addition, the specific names of the various functional modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present disclosure.

[0151] Figure 5 The structure diagram of an electronic device provided by the embodiment of the present disclosure is shown in FIG. Figure 5 As shown, the electronic device 500 includes one or more processors 501 and a memory 502 .

[0152] The processor 501 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 500 to perform desired functions.

[0153] The memory 502 may include one or more computer program products, and the computer program products may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 501 may execute the program instructions to implement the information processing method of the embodiment of the present disclosure described above and / or other desired functions.

[0154] In one example, the electronic device 500 may further include: an input device 503 and an output device 504, which are interconnected via a bus system and / or other forms of connection mechanisms (not shown). The input device 503 may include, for example, a keyboard, a mouse, etc. The output device 504 may output various information to the outside. The output device 504 may include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, etc.

[0155] Of course, to simplify, Figure 5 Only some of the components related to the present disclosure in the electronic device 500 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application situations, the electronic device 500 may also include any other appropriate components.

[0156] In addition to the above-mentioned method and device, the embodiments of the present disclosure may also be a computer program product, which includes computer program instructions. When the computer program instructions are executed by a processor, the processor executes the information processing method provided by the embodiments of the present disclosure.

[0157] The computer program product may be written in any combination of one or more programming languages ​​to write program code for performing the operations of the disclosed embodiments, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0158] In addition, the embodiments of the present disclosure may also be a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the information processing method provided by the embodiments of the present disclosure.

[0159] Computer readable storage media can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable 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 above.

[0160] It should be noted that the terms used in the present disclosure are only for describing specific embodiments, rather than limiting the scope of the present application. As shown in the present disclosure specification and claims, unless the context clearly indicates an exception, the words "one", "a", "a kind of" and / or "the" do not specifically refer to the singular, but may also include the plural. The term "and / or" includes any and all combinations of one or more related listed items. Relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. The term "include", "comprise" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method or device including a series of elements includes not only those elements, but also includes other elements that are not explicitly listed, or also includes elements inherent to such process, method or device. In the absence of more restrictions, the elements defined by the sentence "including one..." do not exclude the existence of other identical elements in the process, method or device including the elements.

[0161] The above description is only a specific embodiment of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An information processing method, characterized in that: include: Obtain the target tasks to be processed and initial associated information; The initial associated information is unprocessed information that is associated with the processing process of the target task; The initial association information is stored in units of the first sentence granularity; The first sentence granularity represents the length of information; Splitting the initial association information to generate a plurality of first association information of a second sentence granularity; The second statement granularity represents the length of information, and the second statement granularity is smaller than the first statement granularity; The first associated information is information with complete semantics; Extracting features from the target task and each of the first associated information, and determining the degree of association between the extracted features as the degree of association between the target task and each of the first associated information; the degree of association is used to characterize the degree of similarity between two pieces of information; Selecting the first association information whose association degree is greater than a first threshold from each of the first association information as the second association information; An information extraction process of content summary is performed on each of the second related information to generate target related information corresponding to the target task.

2. The method according to claim 1, characterized in that After selecting the second association information whose association degree is greater than the first threshold from each of the first association information, the method further includes: Selecting the first association information whose association degree is greater than a second threshold from each of the first association information as the third association information; the second threshold is greater than the first threshold; The step of performing information extraction processing on each of the second associated information to summarize the content and generate target associated information corresponding to the target task includes: performing information extraction processing of content summary on each of the second related information to generate fourth related information; Determine a union of the third association information and the fourth association information as the target association information corresponding to the target task; Alternatively, comparing the difference between the amount of information of the third related information and the amount of information of the fourth related information as the information difference between the third related information and the fourth related information; and / or performing keyword matching on the third related information and the fourth related information, and determining the information difference between the third related information and the fourth related information based on keywords that have not been successfully matched; A weight value is determined based on the information difference, and a weighted fusion process is performed on the third association information and the fourth association information based on the weight value to generate the target association information corresponding to the target task.

3. The method according to claim 1 or 2, characterized in that: The information extraction process of summarizing the content of each of the second related information includes: The degree of association corresponding to the second related information is used as the attention weight, and an information extraction algorithm based on the attention mechanism is used to perform information extraction processing for content summarization on each of the second related information.

4. The method according to claim 1, characterized in that: The sentence-by-sentence processing of the initial association information to generate a plurality of first association information of a second sentence granularity includes: The initial association information is processed based on semantics to generate a plurality of fifth association information of a third sentence granularity; the third sentence granularity represents the length of the information; the third sentence granularity is smaller than the first sentence granularity, and the third sentence granularity is larger than the second sentence granularity; the fifth association information is information with complete semantics; Perform sentence segmentation based on grammar on any of the fifth association information to generate multiple pieces of the first association information at the second sentence granularity.

5. The method according to claim 1, characterized in that After obtaining the target task to be processed and the initial associated information, the method further includes: Analyzing the target task from at least one task dimension, determining an analysis result of the task dimension, and determining the complexity of the target task based on the analysis result of the task dimension; wherein the task dimension is a dimension that reflects the complexity of the task; Analyzing the initial associated information from at least one information volume dimension, determining an analysis result of the information volume dimension, and determining the information volume of the initial associated information based on the analysis result of the information volume dimension; wherein the information volume dimension is a dimension reflecting the amount of information; Based on the complexity of the target task and / or the amount of information of the initial associated information, determine an information processing method for the initial associated information; wherein the information processing method includes retaining the initial associated information, performing information extraction processing of content summarization on the initial associated information, or triggering the execution of the step of performing sentence processing on the initial associated information to generate a plurality of first associated information of a second sentence granularity; The initial association information is processed according to the information processing method to obtain the target association information.

6. The method according to claim 5, characterized in that Based on the complexity of the target task and the amount of information of the initial associated information, determining an information processing method for the initial associated information includes: If the amount of information of the initial association information is less than a third threshold, determining that the information processing method is to retain the initial association information; If the amount of information of the initial associated information is greater than or equal to the third threshold, and the complexity of the target task is less than the fourth threshold, determining that the information processing method is to perform information extraction processing for content summarization on the initial associated information; If the amount of information of the initial associated information is greater than or equal to the third threshold, and the complexity of the target task is greater than or equal to the fourth threshold, then the information processing method is determined to trigger the execution of the step of splitting the initial associated information to generate multiple first associated information of the second statement granularity.

7. The method according to claim 1, characterized in that The obtaining of the target task to be processed includes: Acquire an initial task to be processed, and determine the initial task as the target task; Alternatively, obtaining the target task to be processed includes: Get the initial tasks to be processed; The initial task is split to generate a plurality of subtasks corresponding to the initial task; wherein the task contents of the subtasks have complete semantics; Any of the subtasks is determined as the target task.

8. An information processing device, characterized in that: include: A target task acquisition module is used to acquire the target task to be processed and initial associated information; The initial associated information is unprocessed information that is associated with the processing process of the target task; The initial association information is stored in units of the first sentence granularity; The first sentence granularity represents the length of information; A first correlation information generating module, configured to split the initial correlation information to generate a plurality of first correlation information of a second sentence granularity; The second statement granularity represents the length of information, and the second statement granularity is smaller than the first statement granularity; The first associated information is information with complete semantics; a correlation determination module, used to extract features from the target problem and each of the first correlation information, and determine the correlation between the extracted features as the correlation between the target task and each of the first correlation information; the correlation is used to represent the similarity between the two pieces of information; A second association information screening module, configured to screen the first association information whose association degree is greater than a first threshold from each of the first association information as the second association information; The target-related information generation module is used to perform information extraction processing on the content summary of each of the second related information to generate target-related information corresponding to the target task.

9. An electronic device, characterized in that: The electronic device comprises: Processor and memory; The processor is used to execute the information processing method according to any one of claims 1 to 7 by calling the program or instruction stored in the memory.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program or an instruction, wherein the program or the instruction causes a computer to execute the information processing method according to any one of claims 1 to 7.

Images