CN113535899A - Automatic studying and judging method for internet information emotion tendentiousness - Google Patents
Automatic studying and judging method for internet information emotion tendentiousness Download PDFInfo
- Publication number
- CN113535899A CN113535899A CN202110769546.3A CN202110769546A CN113535899A CN 113535899 A CN113535899 A CN 113535899A CN 202110769546 A CN202110769546 A CN 202110769546A CN 113535899 A CN113535899 A CN 113535899A
- Authority
- CN
- China
- Prior art keywords
- training
- model
- data
- public sentiment
- data set
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 230000008451 emotion Effects 0.000 title claims abstract description 26
- 238000012549 training Methods 0.000 claims abstract description 51
- 238000011160 research Methods 0.000 claims abstract description 7
- 238000013135 deep learning Methods 0.000 claims abstract description 5
- 230000002996 emotional effect Effects 0.000 claims description 21
- 238000007781 pre-processing Methods 0.000 claims description 8
- 230000007935 neutral effect Effects 0.000 claims description 7
- 230000003068 static effect Effects 0.000 claims description 5
- 230000005540 biological transmission Effects 0.000 claims description 3
- 238000004140 cleaning Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000000547 structure data Methods 0.000 description 2
- 241000271566 Aves Species 0.000 description 1
- 241000272194 Ciconiiformes Species 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000008267 milk Substances 0.000 description 1
- 210000004080 milk Anatomy 0.000 description 1
- 235000013336 milk Nutrition 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/374—Thesaurus
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/211—Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/253—Grammatical analysis; Style critique
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Databases & Information Systems (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Machine Translation (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses an automatic studying and judging method for internet information emotion tendentiousness, relates to the technical field of language emotion analysis, and adopts a method of pre-training by using a RoBERTA model on general linguistic data and fine-tuning a downstream task. And a mixed precision and multi-machine multi-GPU training mode is used in the deep learning training process. After the super-participation training is found, a model is deployed and an interface is provided to finish the automatic research and judgment work; the method solves the problems of low accuracy, poor generalization effect of a research and judgment model, poor performance in dealing with complex Chinese contexts such as tarnish, ambiguity and the like in the traditional public sentiment emotion research and judgment work.
Description
Technical Field
The invention relates to the technical field of language emotion analysis, in particular to an automatic studying and judging method for emotion tendencies of internet information.
Background
According to the display of 'Chinese Internet development statistics report' of 47 th phase issued by a Chinese Internet information center (CNNIC), the number of Chinese Internet users reaches 9.89 hundred million by 12 months and 20 days in 2020. Therefore, the internet provides a great deal of data information to us, wherein analysis of the netizen public opinion is an essential step for the network public opinion analysis.
With the continuous and deep development of the internet era, the internet public opinion sentiment analysis has become an indispensable means for understanding the social opinion, holding the public opinion trend, and quickly responding and processing the emergencies. The automatic research and judgment of the emotional tendency of Internet public sentiment is a vivid application combining big data and artificial intelligence.
However, the existing emotion analysis scheme generally adopts the technologies of traditional machine learning, support vector machine, logistic regression, CNN neural network, LSTM neural network and the like, and the existing technology is not good enough in terms of natural language processing, which is the core of automatic study and judgment of emotion tendentiousness, when facing complex Chinese contexts such as vagueness, ambiguity and the like, the model generalization effect is poor, and the emotion tendentiousness judgment accuracy rate has a great space for improvement.
Aiming at the problems in the prior art, the application provides an automatic studying and judging method for the emotion tendentiousness of internet information, and solves the problems that the accuracy rate is low, the generalization effect of a studying and judging model is not good enough, and the performance is poor when complex Chinese contexts such as vague and ambiguous are dealt with in the traditional public opinion emotion studying and judging work.
Disclosure of Invention
The invention aims to provide an automatic studying and judging method aiming at internet information emotion tendentiousness, and solves the problems of low accuracy, poor generalization effect of studying and judging models and poor performance in dealing with complex Chinese contexts such as vague and ambiguous and the like in the traditional public opinion emotion studying and judging work.
The invention provides an automatic studying and judging method for internet information emotion tendentiousness, which comprises the following steps:
establishing a public opinion corpus data set;
establishing a RoBERTA model, importing public sentiment corpus data sets for pre-training, improving the Bert of the RoBERTA model, and obtaining a pre-training model;
fine-tuning parameters of a pre-training model based on a downstream task data set, and storing a final model after fine tuning;
and after the final model is used for prediction, the emotional tendency probability is output, and automatic research and judgment are realized.
Further, the public sentiment corpus data set is preprocessed, and the preprocessing steps are as follows:
public sentiment data of emotion tendentiousness marked in a public sentiment corpus data set are collected, and data cleaning is carried out on the data;
formatting public opinion data;
converting public opinion data as required by using a Chinese character dictionary file;
and carrying out multi-process preprocessing on the public sentiment data.
Furthermore, pre-training of the public sentiment corpus data set is performed based on deep learning, and a mixed precision multi-machine multi-GPU training mode is used in the training process.
Further, the pre-training of the public sentiment corpus data set includes improving Bert, and specifically includes:
removing the NSP task;
specifying a BERT mask type;
the static Mask varies the state Mask.
Further, the learning rate parameter of the fine tuning pre-training model is 3e-4, the batch size parameter is 64, the epochs parameter is 12, and the mask type is set to full _ visible.
Further, the final model configures an HTTP interface, the data submission mode adopted by the HTTP interface is POST, and the transmission format is JSON.
Further, the emotional tendency output after model prediction comprises positive emotional tendency, negative emotional tendency, neutral emotional tendency and irrelevant emotional tendency.
Compared with the prior art, the invention has the following remarkable advantages:
the invention provides an automatic studying and judging method for internet information emotion tendentiousness, which adopts a method of pre-training by using a RoBERTA model on general linguistic data and finely adjusting downstream tasks. And a mixed precision and multi-machine multi-GPU training mode is used in the deep learning training process. After the super-participatory training is found, the model is deployed and an interface is provided to finish the automatic studying and judging work, and the method is good in robustness and strong in model generalization capability, and can provide the studying and judging result with high accuracy in the face of special Chinese context.
The invention provides an automatic studying and judging method for internet information emotion tendentiousness, which improves Bert in a pre-training process, changes static Mask into dynamic Mask, indirectly increases training data and is beneficial to improving model performance. Eliminating NSP loss can be on the same level or slightly improved in the performance of downstream tasks as the original BERT.
And thirdly, the RoBERTA model (160G) uses data which is 10 times more than that of the Bert model (16G) in the automatic judging method for the internet information emotion tendentiousness. More training data increases the diversity of the vocabulary, syntactic structure, and syntactic structure data.
Drawings
FIG. 1 is a block diagram of a fine-tuning pre-training model according to an embodiment of the present invention;
FIG. 2 is a diagram of model architecture differences before training provided by an embodiment of the present invention;
fig. 3 is a schematic diagram illustrating a fine adjustment principle of RoBERTa on different tasks according to an embodiment of the present invention;
fig. 4 is a fine-tuned MNLI accuracy diagram provided by an embodiment of the present invention;
fig. 5 is a diagram illustrating a BERT input portion according to an embodiment of the present invention.
Detailed Description
The technical solutions of the embodiments of the present invention are clearly and completely described below with reference to the drawings in the present invention, and it is obvious that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, shall fall within the scope of protection of the present invention.
Referring to fig. 1-5, the invention provides an automatic studying and judging method for internet information emotion tendentiousness, which comprises the following steps:
establishing a public opinion corpus data set, and preprocessing the public opinion corpus data set;
establishing a RoBERTA model, designating a target task of the model, importing a preprocessed public sentiment corpus data set for pre-training, improving the Bert of the RoBERTA model, and obtaining a pre-training model;
fine-tuning parameters of a pre-training model based on a downstream task data set, and storing a final model after fine tuning;
and after the final model is used for prediction, the emotional tendency probability is output, and automatic research and judgment are realized.
And the final model is configured with an HTTP interface, the data submission mode adopted by the HTTP interface is POST, the transmission format is JSON, and the emotional tendency probability is output after model prediction, so that automatic study and judgment are realized.
The emotional tendency output after model prediction comprises positive emotional tendency, negative emotional tendency, neutral emotional tendency and irrelevant emotional tendency.
Example 1
The method comprises the following steps of preprocessing a public opinion corpus data set, wherein the preprocessing comprises the following steps:
public sentiment data of emotion tendentiousness marked in a public sentiment corpus data set are collected, and data cleaning is carried out on the data;
formatting public opinion data;
converting public opinion data as required by using a Chinese character dictionary file;
and carrying out multi-process preprocessing on the public sentiment data.
Example 2
Referring to fig. 2, the differences between the model architectures before training: BERT uses a bidirectional transducer. OpenAI GPT uses a left to right transform. ELMo uses independently trained left-to-right and right-to-left LSTMs concatenations to generate features for downstream tasks. Of these three representations, the BERT representation alone is a condition for the joint representation to have both left and right contexts on all layers. In addition to architectural differences, BERT and OpenAI GPT are fine-tuning methods, while ELMo is a feature-based method.
The pre-training of the public opinion corpus data set is carried out based on deep learning, and a mixed precision multi-machine multi-GPU training mode is used in the training process. The application uses 160G training public sentiment corpora, uses RoBERTA model training a week, the batch size is 64, and 3 machines each machine of 8 GPU (NVIDIA Tesla V10016G) carry out pre-training totally.
Pre-training public opinion corpus data sets includes improving Bert, specifically including:
removing the NSP task;
specifying a BERT mask type;
the static Mask varies the state Mask.
The RoBERTA improves the Bert on the training method, and mainly embodies four aspects of changing the mask mode, discarding NSP tasks, training over-parameter optimization and using larger-scale training data. The improvement points are as follows: (1) static Mask shift Mask: the dynamic mask is equivalent to indirectly increasing training data, and is beneficial to improving the model performance; (2) NSP task removal: in order to capture the relationship between sentences, Bert uses the NSP task for pre-training, namely, a pair of sentences A and B is input, whether the two sentences are continuous or not is judged, the sum of the maximum lengths of the two sentences is 512, and after removing NSP by RoBERTA, a plurality of continuous sentences are input each time until the maximum length is 512 (and can cross articles). Eliminating NSP loss can be on the same level or slightly improved in the performance of downstream tasks as the original BERT. Because the Bert single sentence is input as a unit, the model cannot learn the remote dependency relationship among the words, and the RoBERTA input is a plurality of continuous sentences, the model can capture longer dependency relationship, which is friendly to the downstream task of a long sequence; (3) larger batch size: the batch size of RoBERTA is 8 k. By using the training strategy in machine translation for reference, the phenomenon that the optimization rate and the model performance can be improved by matching a larger batch size with a larger learning rate is adopted, and experiments prove that the actually Bert can also use the larger batch size. (4) More training data: for longer training times, RoBERTA (160G) uses 10 times more data than Bert (16G). More training data increases the diversity of the vocabulary, syntactic structure, and syntactic structure data.
Example 3
As shown in FIGS. 1 and 3, the learning rate parameter of the fine-tuned pre-trained model on the downstream task data set is 3e-4, the batch size parameter is 64, the epochs parameter is 12, and the mask type is set to full _ visible.
The RoBERTa model performs comprehensive pre-training and fine-tuning of BERT. The same architecture is used in both pre-training and fine-tuning, except for the output layer. The same pre-trained model parameters are used to initialize the models for different downstream tasks. During the fine tuning, all parameters are fine tuned. [ CLS ] is a special symbol added before each input instance, [ SEP ] is a special delimiter mark to separate the questions/answers.
The prediction task of the final model of the application is as follows:
Input＝[CLS]the man went to[MASK]store[SEP]he bought a gallon [MASK]milk[SEP]
Label＝IsNext
Input＝[CLS]the man[MASK]to the store[SEP]penguin[MASK]are flight ##less birds[SEP]
Label＝NotNext
as in fig. 4, the MNLI accuracy after fine tuning is shown, starting with model parameters previously trained for k steps. The x-axis is the value of k.
Example 4
Taking the example of neutral news, its BERT input section is shown in fig. 5, where the BERT input representation illustrates that the input embedding is the sum of mark embedding, segment embedding, and position embedding.
The model of the application predicts that the result is as follows:
{"label":"__label__neutral","probability":0.99930811524391174}
wherein, __ label __ neutral represents that the model predicts neutral emotion, the probability is 99%, and the news is judged to be neutral news.
Taking negative news as an example, the result is as follows through model prediction:
{"label":"__label__negative","probability":0.9227299332618713}
wherein "__ label __ negative" indicates that the model predicts negative emotion, the probability is 92%, and the news is judged to be negative.
Taking the positive news as an example, the result is predicted by a model as follows:
{"label":"__label__positive","probability":0.9998956918716431}
wherein "__ label __ positive" indicates that the model predicts positive emotion and the probability is 99%, and the model is judged as positive news.
In the practical application scene, the public sentiment events are complex and changeable, the public sentiment focus is not single, the public sentiment progresses in multiple levels, and the method can still accurately judge the sentiment tendency of the public sentiment text.
The above disclosure is only for a few specific embodiments of the present invention, however, the present invention is not limited to the above embodiments, and any variations that can be made by those skilled in the art are intended to fall within the scope of the present invention.
Claims (7)
1. An automatic studying and judging method for internet information emotion tendentiousness is characterized by comprising the following steps:
establishing a public opinion corpus data set;
establishing a RoBERTA model, importing public sentiment corpus data sets for pre-training, improving the Bert of the RoBERTA model, and obtaining a pre-training model;
fine-tuning parameters of a pre-training model based on a downstream task data set, and storing a final model after fine tuning;
and after the final model is used for prediction, the emotional tendency probability is output, and automatic research and judgment are realized.
2. The method as claimed in claim 1, wherein the public sentiment corpus data set is preprocessed, the preprocessing comprises:
public sentiment data of emotion tendentiousness marked in a public sentiment corpus data set are collected, and data cleaning is carried out on the data;
formatting public opinion data;
converting public opinion data as required by using a Chinese character dictionary file;
and carrying out multi-process preprocessing on the public sentiment data.
3. The method as claimed in claim 1, wherein the pre-training of the public sentiment corpus data set is based on deep learning, and a mixed-precision, multi-machine and multi-GPU training mode is used in the training process.
4. The method as claimed in claim 3, wherein the pre-training of the corpus data set includes Bert improvement, and specifically includes:
removing the NSP task;
specifying a BERT mask type;
the static Mask varies the state Mask.
5. The method as claimed in claim 1, wherein the learning rate parameter of the fine tuning pre-training model is 3e-4, the batch size parameter is 64, the epochs parameter is 12, and the mask type is set to full _ visible.
6. The method for automatically studying and judging the emotional tendency of the internet information as claimed in claim 1, wherein the final model is configured with an HTTP interface, the HTTP interface adopts a POST data submission mode and a JSON transmission format.
7. The method as claimed in claim 1, wherein the emotional tendencies output after model prediction include positive emotional tendencies, negative emotional tendencies, neutral emotional tendencies and irrelevant emotional tendencies.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110769546.3A CN113535899B (en) | 2021-07-07 | 2021-07-07 | Automatic studying and judging method for emotion tendencies of internet information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110769546.3A CN113535899B (en) | 2021-07-07 | 2021-07-07 | Automatic studying and judging method for emotion tendencies of internet information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113535899A true CN113535899A (en) | 2021-10-22 |
| CN113535899B CN113535899B (en) | 2024-02-27 |
Family
ID=78127065
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110769546.3A Active CN113535899B (en) | 2021-07-07 | 2021-07-07 | Automatic studying and judging method for emotion tendencies of internet information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113535899B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117076751A (en) * | 2023-10-10 | 2023-11-17 | 西安康奈网络科技有限公司 | Public opinion event development trend judging system based on multidimensional feature analysis |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017107957A1 (en) * | 2015-12-22 | 2017-06-29 | 中兴通讯股份有限公司 | Human face image retrieval method and apparatus |
| CN111667069A (en) * | 2020-06-10 | 2020-09-15 | 中国工商银行股份有限公司 | Pre-training model compression method and device and electronic equipment |
| CN112183486A (en) * | 2020-11-02 | 2021-01-05 | 中山大学 | Method for rapidly identifying single-molecule nanopore sequencing base based on deep network |
| CN112270181A (en) * | 2020-11-03 | 2021-01-26 | 北京明略软件系统有限公司 | Sequence labeling method, system, computer readable storage medium and computer device |
| CN112434523A (en) * | 2020-11-25 | 2021-03-02 | 上海极链网络科技有限公司 | Text auditing device and method for reducing false alarm rate of harmonic matching of sensitive words |
| CN112508609A (en) * | 2020-12-07 | 2021-03-16 | 深圳市欢太科技有限公司 | Crowd expansion prediction method, device, equipment and storage medium |
| CN112529146A (en) * | 2019-09-18 | 2021-03-19 | 华为技术有限公司 | Method and device for training neural network model |
| CN112597759A (en) * | 2020-11-30 | 2021-04-02 | 深延科技(北京)有限公司 | Text-based emotion detection method and device, computer equipment and medium |
| CN112784041A (en) * | 2021-01-06 | 2021-05-11 | 河海大学 | Chinese short text emotion orientation analysis method |
| CN112883720A (en) * | 2021-01-25 | 2021-06-01 | 北京瑞友科技股份有限公司 | Text emotion classification system and method based on double models |
| CN113011185A (en) * | 2020-07-17 | 2021-06-22 | 上海浦东华宇信息技术有限公司 | Legal field text analysis and identification method, system, storage medium and terminal |
-
2021
- 2021-07-07 CN CN202110769546.3A patent/CN113535899B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017107957A1 (en) * | 2015-12-22 | 2017-06-29 | 中兴通讯股份有限公司 | Human face image retrieval method and apparatus |
| CN112529146A (en) * | 2019-09-18 | 2021-03-19 | 华为技术有限公司 | Method and device for training neural network model |
| CN111667069A (en) * | 2020-06-10 | 2020-09-15 | 中国工商银行股份有限公司 | Pre-training model compression method and device and electronic equipment |
| CN113011185A (en) * | 2020-07-17 | 2021-06-22 | 上海浦东华宇信息技术有限公司 | Legal field text analysis and identification method, system, storage medium and terminal |
| CN112183486A (en) * | 2020-11-02 | 2021-01-05 | 中山大学 | Method for rapidly identifying single-molecule nanopore sequencing base based on deep network |
| CN112270181A (en) * | 2020-11-03 | 2021-01-26 | 北京明略软件系统有限公司 | Sequence labeling method, system, computer readable storage medium and computer device |
| CN112434523A (en) * | 2020-11-25 | 2021-03-02 | 上海极链网络科技有限公司 | Text auditing device and method for reducing false alarm rate of harmonic matching of sensitive words |
| CN112597759A (en) * | 2020-11-30 | 2021-04-02 | 深延科技(北京)有限公司 | Text-based emotion detection method and device, computer equipment and medium |
| CN112508609A (en) * | 2020-12-07 | 2021-03-16 | 深圳市欢太科技有限公司 | Crowd expansion prediction method, device, equipment and storage medium |
| CN112784041A (en) * | 2021-01-06 | 2021-05-11 | 河海大学 | Chinese short text emotion orientation analysis method |
| CN112883720A (en) * | 2021-01-25 | 2021-06-01 | 北京瑞友科技股份有限公司 | Text emotion classification system and method based on double models |
Non-Patent Citations (2)
| Title |
|---|
| 徐菲菲;冯东升;: "文本词向量与预训练语言模型研究", 上海电力大学学报, no. 04, 15 August 2020 (2020-08-15) * |
| 杜博;万国佳;纪颖;: "基于几何深度学习的知识图谱关键技术研究进展", 航空兵器, no. 03, 28 May 2020 (2020-05-28) * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117076751A (en) * | 2023-10-10 | 2023-11-17 | 西安康奈网络科技有限公司 | Public opinion event development trend judging system based on multidimensional feature analysis |
| CN117076751B (en) * | 2023-10-10 | 2024-01-16 | 西安康奈网络科技有限公司 | Public opinion event development trend judging system based on multidimensional feature analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113535899B (en) | 2024-02-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7127106B2 (en) | Question answering process, language model training method, apparatus, equipment and storage medium | |
| KR102350543B1 (en) | Semantic representation model processing method, device, electronic equipment and storage medium | |
| CN110309839B (en) | A kind of method and device of iamge description | |
| CN111192692B (en) | Entity relationship determination method and device, electronic equipment and storage medium | |
| CN110134946B (en) | Machine reading understanding method for complex data | |
| CN110929030A (en) | Text abstract and emotion classification combined training method | |
| CN109117485B (en) | Method and device for generating blessing language text and computer readable storage medium | |
| CN110532554A (en) | A kind of Chinese abstraction generating method, system and storage medium | |
| CN107392147A (en) | A kind of image sentence conversion method based on improved production confrontation network | |
| CN112699216A (en) | End-to-end language model pre-training method, system, device and storage medium | |
| CN111126061A (en) | Method and device for generating antithetical couplet information | |
| CN113051368B (en) | Double-tower model training method, retrieval device and electronic equipment | |
| CN115392259A (en) | Microblog text sentiment analysis method and system based on confrontation training fusion BERT | |
| CN113032541A (en) | Answer extraction method based on bert and fusion sentence cluster retrieval | |
| CN115294427A (en) | Stylized image description generation method based on transfer learning | |
| CN108763355A (en) | A kind of intelligent robot interaction data processing system and method based on user | |
| CN111626041A (en) | Music comment generation method based on deep learning | |
| CN113535899A (en) | Automatic studying and judging method for internet information emotion tendentiousness | |
| CN113326367B (en) | Task type dialogue method and system based on end-to-end text generation | |
| CN113934835A (en) | Retrieval type reply dialogue method and system combining keywords and semantic understanding representation | |
| CN116522165B (en) | Public opinion text matching system and method based on twin structure | |
| CN110334204B (en) | Exercise similarity calculation recommendation method based on user records | |
| CN112349294A (en) | Voice processing method and device, computer readable medium and electronic equipment | |
| Guo et al. | Ernie-bilstm based Chinese text sentiment classification method | |
| Li et al. | Multilingual toxic text classification model based on deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |