CN-121350255-B - Electronic equipment and traffic data tag generation method

CN121350255BCN 121350255 BCN121350255 BCN 121350255BCN-121350255-B

Abstract

The invention relates to electronic equipment and a traffic data label generation method, wherein the electronic equipment comprises a controller, a label generation model and a judgment unit, wherein the controller is configured to perform blocking processing on traffic public opinion text data to obtain a plurality of data blocks, at least one data block in the plurality of data blocks and prompt word information are input into the label information generation model to obtain at least one information label result, the prompt word is used for indicating the label information generation model to determine the information label result of a traffic event corresponding to the data block through semantic understanding and analysis on the data block, and the information label result comprises labels corresponding to the traffic event and intersection identifications of intersections.

Inventors

GAO HONGJING
DUAN TAO
ZENG JIE
CHEN ZHONG
TAN LEI
WANG CHENGLONG

Assignees

青岛海信网络科技股份有限公司

Dates

Publication Date: 20260508
Application Date: 20251216

Claims (10)

1. An electronic device, comprising: A controller configured to: carrying out block processing on the traffic public opinion text data to obtain a plurality of data blocks; Inputting at least one data block in the plurality of data blocks and prompt word information into a marking information generation model to obtain at least one information marking result, wherein the prompt word is used for indicating the marking information generation model to determine the information marking result of the traffic event corresponding to the data block through semantic understanding and analysis of the data block, and the information marking result comprises a label corresponding to the traffic event and an intersection identifier of an intersection; Acquiring a road topology network corresponding to a target intersection identifier in a first information labeling result and propagation time information between adjacent intersections corresponding to the road topology network, wherein the first information labeling result is one of the at least one information labeling result, and the propagation time information comprises at least one of the following propagation time between the adjacent intersections in different time periods and the propagation time between the adjacent intersections in different road conditions; Generating early warning information based on the road topology network and the propagation time information, wherein the early warning information comprises an early warning intersection identifier, an early warning label and expected arrival time, wherein the early warning intersection indicated by the early warning intersection identifier and the target intersection indicated by the target intersection identifier belong to the same transmission path in the road topology network, the early warning label is determined based on a first label in the first information labeling result, and the expected arrival time is determined based on the transmission time of the target intersection and the early warning intersection in the propagation time information; the early warning information further comprises influence degree information, wherein the influence degree information is used for indicating the influence degree of the traffic event corresponding to the first information labeling result on the early warning intersection; Wherein the influence degree information is determined according to the following formula: ; Wherein, the Predicted arrival at The degree of influence of the individual downstream intersections, The initial influence degree of the target intersection, The number of road segments passing from the target intersection to the early warning intersection, The natural constant is adopted in the method, Decay time constant; maximum attenuation coefficient, 0< lambda less than or equal to 1, Representing the ratio of maximum intensity that affects the retention after infinity, The final range for controlling the impact is associated with the severity of the corresponding traffic event itself.
2. The electronic device of claim 1, wherein the controller is specifically configured to: Obtaining word segmentation results of the traffic public opinion text data, wherein the word segmentation results comprise a plurality of words, word parts corresponding to the words and position information of the words in the traffic public opinion text data; based on the word segmentation result, obtaining a clause result corresponding to the traffic public opinion text data, wherein the clause result comprises a plurality of clauses and position information of the clauses in the traffic public opinion text data; and based on the clause result, obtaining a blocking result of the traffic public opinion text data, wherein the blocking result comprises the plurality of data blocks and position information of the data blocks in the traffic public opinion text data.
3. The electronic device of claim 1, wherein the controller is specifically configured to: Distributing the plurality of data blocks to a plurality of working threads based on a load balancing principle; and calling the annotation information generation model based on the plurality of working threads in parallel, and carrying out semantic understanding and analysis based on the input data block and the prompt word information through the annotation information generation model to obtain the at least one information annotation result.
4. The electronic device of claim 1, wherein the information labeling result further comprises target information for the traffic event, the controller further configured to: filtering the at least one information labeling result based on the target information and the filtering rule to obtain at least one final information labeling result; in the case that the target information comprises the recognition confidence coefficient of the traffic event, the filtering rule comprises filtering out an information labeling result that the recognition confidence coefficient is smaller than or equal to a first confidence coefficient threshold value; when the target information comprises the time stamp of the traffic event, the filtering rule comprises filtering out an information marking result of the time stamp before a preset duration; when the target information comprises evidence texts corresponding to the traffic events, the filtering rules comprise a plurality of information labeling results with association relations between the merged evidence texts and matched labels and intersection identifications; in the case that the target information includes a time stamp and source information, the filtering rule includes merging a plurality of information labeling results that are matched with the time stamp, the tag and the intersection identifier and have different sources.
5. The electronic device of claim 1, wherein the hint word information includes a plurality of tags, the tags including tag initial confidence, the controller further configured to: updating the confidence coefficient of a first label based on at least one of a model output confidence coefficient and an attenuation factor and the label initial confidence coefficient to obtain the first confidence coefficient of the first label, wherein the model output confidence coefficient is used for indicating the identification confidence coefficient of the labeling information generation model for identifying the traffic event corresponding to the first label, the attenuation factor is used for indicating the attenuation coefficient of the first label changing along with time, and the first label is any one of the labels; Determining that the first tag is in a state to be verified under the condition that the first confidence coefficient is smaller than or equal to a first confidence coefficient threshold value; In the state to be verified, if the first label is verified to be available, canceling the state to be verified of the first label, and updating the first confidence coefficient to a second confidence coefficient, wherein the second confidence coefficient is larger than the first confidence coefficient threshold value; And deleting the first tag from the plurality of tags if the first tag is verified to be unavailable.
6. The electronic device of any one of claims 1-5, wherein the controller is further configured to: Acquiring at least one feature corresponding to the second information labeling result in the traffic public opinion text data under the condition that a second label in the second information labeling result is verified to be correct; Updating the weight of a first feature associated under the second tag from a first weight to a second weight, the second weight being greater than the first weight, the first feature being one of the at least one feature; Updating the confidence coefficient of the second label from a third confidence coefficient to a fourth confidence coefficient, wherein the fourth confidence coefficient is larger than the third confidence coefficient; under the condition that a third label in a third information labeling result is modified into a fourth label, at least one target feature corresponding to the third information labeling result in the traffic public opinion text data is obtained; Updating the weight of the second feature associated under the third tag from a third weight to a fourth weight, the fourth weight being less than the third weight, the second feature being one of the at least one target feature; Updating the weight of a third feature associated under the fourth tag from a fifth weight to a sixth weight, the sixth weight being less than the fifth weight, the third feature being one of the at least one target feature; updating the confidence level of the third label from a fifth confidence level to a sixth confidence level, wherein the sixth confidence level is smaller than the fifth confidence level; updating the confidence of the fourth label from a seventh confidence to an eighth confidence, wherein the eighth confidence is greater than the seventh confidence.
7. The electronic device of claim 6, wherein the controller is further configured to: updating the confidence coefficient of the tag and the weight of the feature associated with the tag into the corresponding recognition rule of the tag under the condition that the prompt word information also comprises the recognition rule of the traffic event corresponding to the tag; Or alternatively Updating the confidence coefficient of the tag and the weight of the feature associated with the tag into the determination rule under the condition that the prompt word information also comprises the determination rule of the recognition confidence coefficient of the tag; Or alternatively In the case that the hint word information further includes an example sample, updating the tag and the feature associated with the tag into the corresponding example sample of the tag; Or alternatively Determining at least one candidate tag corresponding to the traffic public opinion text data based on the confidence level of the tag and the weight of the feature associated with the tag; and adding the at least one candidate tag to the prompt word information.
8. The electronic device of claim 6, wherein the controller is further configured to: counting the target weight of the target feature under at least one associated label in the target duration; and determining the target feature as a newly added tag under the condition that the target weight is greater than or equal to a weight threshold.
9. The electronic device of claim 6, wherein the controller is further configured to: Counting the co-occurrence support degree and the co-occurrence confidence degree of the target feature and the target tag in the target duration, wherein the co-occurrence support degree is used for indicating the probability of the co-occurrence of the target feature and the target tag, and the co-occurrence confidence degree is used for indicating the probability of the occurrence of the target tag when the target feature occurs; and determining the target feature as a newly added tag when the co-occurrence support is greater than or equal to a support threshold and the co-occurrence confidence is greater than or equal to a second confidence threshold.
10. A traffic data tag generation method, which is applied to an electronic device, comprising: carrying out block processing on the traffic public opinion text data to obtain a plurality of data blocks; Inputting at least one data block in the plurality of data blocks and prompt word information into a marking information generation model to obtain at least one information marking result, wherein the prompt word is used for indicating the marking information generation model to determine the information marking result of the traffic event corresponding to the data block through semantic understanding and analysis of the data block, and the information marking result comprises a label corresponding to the traffic event and an intersection identifier of an intersection; Acquiring a road topology network corresponding to a target intersection identifier in a first information labeling result and propagation time information between adjacent intersections corresponding to the road topology network, wherein the first information labeling result is one of the at least one information labeling result, and the propagation time information comprises at least one of the following propagation time between the adjacent intersections in different time periods and the propagation time between the adjacent intersections in different road conditions; Generating early warning information based on the road topology network and the propagation time information, wherein the early warning information comprises an early warning intersection identifier, an early warning label and expected arrival time, wherein the early warning intersection indicated by the early warning intersection identifier and the target intersection indicated by the target intersection identifier belong to the same transmission path in the road topology network, the early warning label is determined based on a first label in the first information labeling result, and the expected arrival time is determined based on the transmission time of the target intersection and the early warning intersection in the propagation time information; the early warning information further comprises influence degree information, wherein the influence degree information is used for indicating the influence degree of the traffic event corresponding to the first information labeling result on the early warning intersection; Wherein the influence degree information is determined according to the following formula: ; Wherein, the Predicted arrival at The degree of influence of the individual downstream intersections, The initial influence degree of the target intersection, The number of road segments passing from the target intersection to the early warning intersection, The natural constant is adopted in the method, Decay time constant; maximum attenuation coefficient, 0< lambda less than or equal to 1, Representing the ratio of maximum intensity that affects the retention after infinity, The final range for controlling the impact is associated with the severity of the corresponding traffic event itself.

Description

Electronic equipment and traffic data tag generation method Technical Field Embodiments of the present disclosure relate to data processing technology. And more particularly, to an electronic device and a traffic data tag generation method. Background Traffic public opinion is derived from multiple channels such as news, social media, government affair platforms and the like, most of data forms are unstructured texts, information is fragmented, and expression modes are different remarkably. The existing method mainly depends on keyword matching or static rules, and lacks understanding of complex semantics and context relation, so that the label classification accuracy is insufficient. Disclosure of Invention In order to solve the above technical problems or at least partially solve the above technical problems, an embodiment of the present disclosure provides an electronic device and a traffic data tag generating method. In a first aspect, an embodiment of the disclosure provides an electronic device, which includes a controller configured to perform blocking processing on traffic public opinion text data to obtain a plurality of data blocks, and input at least one data block of the plurality of data blocks and prompting word information into a marking information generating model to obtain at least one information marking result, wherein the prompting word is used for indicating the marking information generating model to determine an information marking result of a traffic event corresponding to the data block through semantic understanding and analysis of the data block, and the information marking result includes a label corresponding to the traffic event and an intersection identifier of an intersection. In the embodiment of the disclosure, firstly, traffic public opinion text data is segmented, then, the data blocks obtained by the segmentation and prompt word information are input into a label information generation model to obtain at least one information label result, so that the text data length can be reduced through segmentation processing, the processing success rate and the processing efficiency are improved, unstructured text data can be effectively understood by understanding and analyzing the data blocks based on the prompt word information through the label information generation model, the structured information label result is obtained by fragmenting the information and the text data with obvious expression mode difference, and the accuracy rate of the information label result can be improved through understanding complex semantics and context relation, so that the utilization rate of the information label result is improved. In some embodiments of the disclosure, the controller is specifically configured to obtain a word segmentation result of the traffic public opinion text data, where the word segmentation result includes a plurality of words, word parts corresponding to the words, and location information of the words in the traffic public opinion text data, obtain a sentence segmentation result corresponding to the traffic public opinion text data based on the word segmentation result, where the sentence segmentation result includes a plurality of sentences, and location information of the sentences in the traffic public opinion text data, and obtain a block result of the traffic public opinion text data based on the sentence segmentation result, where the block result includes the plurality of data blocks, and location information of the data blocks in the traffic public opinion text data. In the embodiment of the disclosure, the obtained data blocks can be finer and more accurate through word segmentation, sentence segmentation and block segmentation in sequence. In some embodiments of the present disclosure, the controller is specifically configured to allocate the plurality of data blocks to a plurality of working threads based on a load balancing principle, call the labeling information generation model in parallel based on the plurality of working threads, and perform speech understanding and analysis based on the input data blocks and the prompt word information through the labeling information generation model to obtain the at least one information labeling result. In the embodiment of the disclosure, the plurality of data blocks are distributed to a plurality of working threads based on a load balancing principle, and then the marking information generation model is called in parallel based on the plurality of working threads to mark the information of the traffic public opinion text data, so that the information marking efficiency can be improved. In some embodiments of the disclosure, the information labeling result further comprises target information of the traffic event, the controller is further configured to perform filtering processing on the at least one information labeling result based on the target information and a filtering rule to obtain at least one final in