CN112487796B - Method and device for sequence labeling and electronic equipment - Google Patents
Method and device for sequence labeling and electronic equipment Download PDFInfo
- Publication number
- CN112487796B CN112487796B CN202011351553.3A CN202011351553A CN112487796B CN 112487796 B CN112487796 B CN 112487796B CN 202011351553 A CN202011351553 A CN 202011351553A CN 112487796 B CN112487796 B CN 112487796B
- Authority
- CN
- China
- Prior art keywords
- feature
- local context
- word vector
- feature map
- sentence
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/40—Processing or translation of natural language
- G06F40/58—Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Abstract
The invention discloses a method and a device for sequence labeling and electronic equipment. According to the method, local context characteristics of a sentence are extracted according to a word vector obtained by converting the sentence, attention expression is obtained according to the word vector and the local context characteristics, and finally the attention expression is input into a sequence labeling layer to obtain a model for sequence labeling. By enhancing the characteristic representation of the local context correlation, the obtained model for sequence labeling can more effectively learn the local context information of the text, so that an accurate recognition result is obtained, and a sequence labeling task and more natural language processing tasks derived from the sequence labeling task are better completed.
Description
Technical Field
The present invention relates to the field of natural language processing technologies, and in particular, to a method and an apparatus for sequence annotation, and an electronic device.
Background
With the development of artificial intelligence technologies such as deep learning, text mining technology brings increasingly profound influence to the industry, wherein more and more services and applications rely on the assistance of technologies such as knowledge extraction to provide better services.
At present, representative models commonly used for sequence labeling include an LSTM model or an attention model, wherein the LSTM model has low computational efficiency, the attention model greatly improves the computational efficiency through an input parallel mode, but because each input is independent, the disadvantage of insufficient utilization of local context information exists, and a final erroneous recognition result is caused.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides the following technical scheme.
One aspect of the present invention provides a method for sequence annotation, comprising:
converting the sentence into a word vector;
extracting local context characteristics of the sentence according to the word vector;
acquiring attention representation according to the word vector and the local context characteristics;
the attention representation is used as an input training sequence annotation model.
Further, before converting the sentence into a word vector, the method further comprises: the sentence is data-washed and the format of the sentence is converted using the BIO method.
Further, the extracting local context features of the sentence according to the word vector includes:
inputting the word vector into a double-layer convolutional neural network model to obtain a feature map (feature map) representing the local context correlation;
the feature map is converted into a feature representation of local context dependencies.
Further, the inputting the word vector into a two-layer convolutional neural network model to obtain a feature map (feature map) representing local context correlation includes:
based on the selected kernel, performing window sliding from left to right in the sentence direction, and performing two-layer convolution operation on the sentence to obtain a feature map (feature map) representing the local context correlation, wherein the sentence is represented as an L-V matrix, L is the preset sentence length, and V is a word vector at the corresponding position; the feature map is represented as a matrix of N L 'V, N being the number of kernels, L' ≦ L.
Further, the converting the feature map (feature map) into a feature representation of local context dependency includes:
averaging the feature maps (feature maps) represented as a matrix of N L 'V to obtain a vector output of L' V;
the vector is converted into a feature matrix of a specified size, i.e., a feature representation of local context correlation, using a feed forward neural network layer.
Further, the obtaining an attention representation according to the word vector and the local context feature includes:
inputting the word vector into an improved multi-head attention model to calculate multi-head attention, wherein the improved multi-head attention model comprises local context features;
and splicing the multiple attention heads to obtain hidden layer state output as an attention representation.
Further, the sequence labeling layer is a conditional random field layer.
The second aspect of the present invention provides an apparatus for sequence annotation, comprising:
the word vector conversion module is used for converting the sentence into a word vector;
the local context feature extraction module is used for extracting local context features of the sentences according to the word vectors;
an attention representation obtaining module, configured to obtain an attention representation according to the word vector and the local context feature;
and the model generation module is used for inputting the attention representation into the sequence annotation layer to obtain a model for sequence annotation.
A third aspect of the invention provides a memory storing a plurality of instructions for implementing the method described above.
A fourth aspect of the present invention provides an electronic device, comprising a processor and a memory connected to the processor, wherein the memory stores a plurality of instructions, and the instructions are loaded and executed by the processor, so that the processor can execute the method.
The invention has the beneficial effects that: according to the technical scheme provided by the invention, the local context characteristics of the sentence are extracted according to the word vector obtained by converting the sentence, the attention expression is obtained according to the word vector and the local context characteristics, and finally the attention expression is input into a sequence labeling layer to obtain a model for sequence labeling. By enhancing the characteristic representation of the local context correlation, the obtained model for sequence labeling can more effectively learn the local context information of the text, so that an accurate recognition result is obtained, and a sequence labeling task and more natural language processing tasks derived from the sequence labeling task are better completed.
Drawings
FIG. 1 is a schematic flow chart of a method for sequence tagging according to the present invention;
FIG. 2 is a schematic structural diagram of a model for sequence annotation according to the present invention;
FIG. 3 is a schematic structural diagram of an apparatus for sequence labeling according to the present invention.
Detailed Description
In order to better understand the technical solution, the technical solution will be described in detail with reference to the drawings and the specific embodiments.
The method provided by the invention can be implemented in the following terminal environment, and the terminal can comprise one or more of the following components: a processor, a memory, and a display screen. Wherein the memory has stored therein at least one instruction that is loaded and executed by the processor to implement the methods described in the embodiments described below.
A processor may include one or more processing cores. The processor connects various parts within the overall terminal using various interfaces and lines, performs various functions of the terminal and processes data by executing or executing instructions, programs, code sets, or instruction sets stored in the memory, and calling data stored in the memory.
The Memory may include a Random Access Memory (RAM) or a Read-Only Memory (ROM). The memory may be used to store instructions, programs, code sets, or instructions.
The display screen is used for displaying user interfaces of all the application programs.
In addition, those skilled in the art will appreciate that the above-described terminal configurations are not intended to be limiting, and that the terminal may include more or fewer components, or some components may be combined, or a different arrangement of components. For example, the terminal further includes a radio frequency circuit, an input unit, a sensor, an audio circuit, a power supply, and other components, which are not described herein again.
Example one
As shown in fig. 1, an embodiment of the present invention provides a method for sequence annotation, including:
s101, converting a sentence into a word vector;
s102, extracting local context characteristics of the sentence according to the word vector;
s103, acquiring attention representation according to the word vector and the local context characteristics;
and S104, inputting the attention representation into a sequence labeling layer to obtain a model for sequence labeling.
In step S101, in order to avoid introducing wrong word boundary information due to word segmentation errors, the embodiment of the present invention represents semantics included in a text based on a word vector model (chinese processing granularity is word level, and english is word level). Pre-training the text to obtain a set X of expression vector of each word:
X＝{x0,x1…xn}
before step S101, the method further includes: the sentence is data-washed and the format of the sentence is converted using the BIO method. Different methods can be used for data cleaning according to specific text formats. For example, for unformatted data, processing of unicode, removing noise data such as meaningless special symbols, and the like may be performed. When the data is cleaned, paragraphs or sentences can be segmented according to actual needs.
In this embodiment, the starting position of a sequence to be labeled can be represented in other ways, such as B start, I continuous, and O, according to the requirement of the sequence labeling specific task, and a category identifier can be added according to the requirement. For example, "Washington university is located in Seattle" requires that "Washington university" (an organization name) be identified with ORG and "Seattle" (a place name) be identified with LOC, then "Washington university is located in Seattle" is denoted as "Hua-B-ORG, Sheng-I-ORG, dun-I-ORG, bit-O, at-O, West-B-LOC, Ya-I-LOC, Diagram-I-LOC", the remaining characters are identified as a task-related category, or other character O.
Step S102 is performed, including:
inputting the word vector into a double-layer convolutional neural network model to obtain a feature map (feature map) representing the local context correlation;
the feature map is converted into a feature representation of local context dependencies.
Specifically, the following method can be adopted for implementation:
firstly, based on a selected kernel, performing window sliding from left to right in the sentence direction, and performing two-layer convolution operation on a sentence to obtain a feature map (feature map) representing local context correlation, wherein the sentence is represented as an L-V matrix, L is a preset sentence length, and V is a word vector at a corresponding position; feature maps (feature maps) are represented as a matrix of N L 'V, N being the number of kernels, L' ≦ L;
then, averaging the feature maps (feature maps) of the matrix expressed as N L 'V to obtain a vector output of L' V;
the vector is converted into a feature matrix of a specified size, i.e., a feature representation of local context correlation, using a feed forward neural network layer.
Step S103 is performed, including:
inputting the word vector into an improved multi-head attention model to calculate multi-head attention, wherein the improved multi-head attention model comprises local context features;
and splicing the multiple attention heads to obtain hidden layer state output as an attention representation.
Inputting the word vector into an improved multi-head attention model, and calculating the multi-head attention by adopting the following formula:
wherein, FlocalIn the form of a local context-feature matrix,
calculated for word vector input in an unmodified multi-head attention modelAttention is paid to the multiple heads.
By the method, the local context features are introduced into the multi-head attention model, so that the attention model can enhance the local context features while acquiring the global context features. Therefore, the model can better learn the text context information so as to more accurately recognize the text.
And step S104 is executed, the attention expression is input into a sequence annotation layer, and a model for sequence annotation is obtained.
In the embodiment of the present invention, the sequence label layer may be a conditional random field layer.
Specifically, the following method can be adopted for implementation:
the attention obtained in step S103 indicates that a Conditional Random Field (CRF) layer is input for sequence type prediction.
Where X (X) is given for a given sentence1,x2,…,xn) Having the predicted tag sequence y ═ (y)1,y2,…,yn) The prediction score can be defined as:
wherein A is a transition score matrix, Ai,jRepresenting the transition score of label i to label j.
Denotes xtMapping to the ytThe score of the marker. PtCan be precisely defined as formula Pt＝Wsht+bsWherein h istHidden layer state output for attention, WsAnd bsAre trainable parameters.
The probability that the sentence X is labeled as the sequence y can be calculated as
Wherein, YXAll possible annotation sequences for a given sentence X
Maximizing annotation sequences using maximum likelihood estimation
Logarithm of likelihood ratio of
Find the highest conditional probability y to serialize the labeled output when decoding:
and finally, outputting a format required by the specific task by using a conversion system.
As an embodiment, fig. 2 is a model for sequence annotation, and the method provided by the present invention can be implemented by the model shown in fig. 2. The model shown in fig. 2 includes: convolutional layers, mapping layers, attention layers, and CRF layers.
The method provided by the embodiment of the invention enhances the characteristic representation of the local context correlation, so that the model for sequence labeling can more effectively learn the local context information of the text, thereby obtaining an accurate identification result.
The method provided by the invention can be applied to all tasks which can be converted into sequence tagging problems, has better effect in the field, can be directly applied to specific tasks of sequence tagging such as word segmentation, part of speech tagging, information extraction and the like through the model for sequence tagging obtained by training of the invention, and can serve more subsequent tasks processed by natural language such as translation, recommendation, reasoning, conversation and the like.
Example two
As shown in fig. 3, another aspect of the present invention further includes a functional module architecture completely corresponding to the foregoing method flow, that is, an embodiment of the present invention further provides an apparatus for sequence tagging, including:
a word vector conversion module 301, configured to convert a sentence into a word vector;
a local context feature extraction module 302, configured to extract local context features of a sentence according to the word vector;
an attention representation obtaining module 303, configured to obtain an attention representation according to the word vector and the local context feature;
and a model generation module 304, configured to input the attention representation into a sequence annotation layer, so as to obtain a model for sequence annotation.
The device provided by the embodiment of the invention also comprises a data preprocessing module which is used for cleaning the sentence and converting the format of the sentence by using a BIO method before converting the sentence into the word vector.
The local context feature extraction module is specifically configured to:
inputting the word vector into a double-layer convolutional neural network model to obtain a feature map (feature map) representing the local context correlation;
the feature map is converted into a feature representation of local context dependencies.
Further, the inputting the word vector into a two-layer convolutional neural network model to obtain a feature map (feature map) representing local context correlation includes:
based on the selected kernel, performing window sliding from left to right in the sentence direction, and performing two-layer convolution operation on the sentence to obtain a feature map (feature map) representing the local context correlation, wherein the sentence is represented as an L-V matrix, L is the preset sentence length, and V is a word vector at the corresponding position; the feature map is represented as a matrix of N L 'V, N being the number of kernels, L' ≦ L.
Further, the converting the feature map (feature map) into a feature representation of local context dependency includes:
averaging the feature maps (feature maps) represented as a matrix of N L 'V to obtain a vector output of L' V;
the vector is converted into a feature matrix of a specified size, i.e., a feature representation of local context correlation, using a feed forward neural network layer.
The attention expression obtaining module is specifically configured to:
inputting the word vector into an improved multi-head attention model to calculate multi-head attention, wherein the improved multi-head attention model comprises local context features;
and splicing the multiple attention heads to obtain hidden layer state output as an attention representation.
In the model generation module, the sequence annotation layer may be a conditional random field layer.
The device can be implemented by the method for sequence annotation provided in the first embodiment, and specific implementation methods can be referred to the description in the first embodiment and are not described herein again.
The invention also provides a memory storing a plurality of instructions for implementing the method according to the first embodiment.
The invention also provides an electronic device comprising a processor and a memory connected to the processor, wherein the memory stores a plurality of instructions, and the instructions can be loaded and executed by the processor to enable the processor to execute the method according to the first embodiment.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention. It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (5)
1. A method for sequence annotation, comprising:
converting the sentence into a word vector;
extracting local context characteristics of the sentence according to the word vector;
acquiring attention representation according to the word vector and the local context characteristics;
inputting the attention representation into a sequence labeling layer to obtain a model for sequence labeling;
before converting the sentence into a word vector, the method further comprises the following steps: washing the sentences and converting the formats of the sentences by using a BIO method;
the extracting of the local context feature of the sentence according to the word vector comprises:
inputting the word vector into a double-layer convolutional neural network model to obtain a feature map (feature map) representing the local context correlation;
-converting the feature map (feature map) into a feature representation of local context dependencies;
inputting the word vector into a double-layer convolutional neural network model to obtain a feature map (feature map) representing local context correlation, including:
based on the selected kernel, performing window sliding from left to right in the sentence direction, and performing two-layer convolution operation on the sentence to obtain a feature map (feature map) representing the local context correlation, wherein the sentence is represented as an L-V matrix, L is the preset sentence length, and V is a word vector at the corresponding position; feature maps (feature maps) are represented as a matrix of N L 'V, N being the number of kernels, L' ≦ L;
said converting said feature map (feature map) into a feature representation of local context dependencies, comprising:
averaging the feature maps (feature maps) represented as a matrix of N L 'V to obtain a vector output of L' V;
converting the vector into a feature matrix with a specified size, namely a feature representation of local context correlation by utilizing a feedforward neural network layer;
the obtaining an attention representation according to the word vector and the local context feature includes:
inputting the word vector into an improved multi-head attention model to calculate multi-head attention, wherein the improved multi-head attention model comprises local context features;
and splicing the multiple attention heads to obtain hidden layer state output as an attention representation.
2. The method for sequence annotation of claim 1, wherein the sequence annotation layer is a conditional random field layer.
3. An apparatus for sequence annotation, comprising:
the word vector conversion module is used for converting the sentence into a word vector;
the local context feature extraction module is used for extracting local context features of the sentences according to the word vectors;
an attention representation obtaining module, configured to obtain an attention representation according to the word vector and the local context feature;
the model generation module is used for inputting the attention representation into a sequence labeling layer to obtain a model for sequence labeling;
the system also comprises a data preprocessing module, a word vector generating module and a word vector generating module, wherein the data preprocessing module is used for cleaning sentences and converting the formats of the sentences by using a BIO method before converting the sentences into word vectors;
the local context feature extraction module is specifically configured to:
inputting the word vector into a double-layer convolutional neural network model to obtain a feature map (feature map) representing the local context correlation;
-converting the feature map (feature map) into a feature representation of local context dependencies;
inputting the word vector into a double-layer convolutional neural network model to obtain a feature map (feature map) representing local context correlation, including:
based on the selected kernel, performing window sliding from left to right in the sentence direction, and performing two-layer convolution operation on the sentence to obtain a feature map (feature map) representing the local context correlation, wherein the sentence is represented as an L-V matrix, L is the preset sentence length, and V is a word vector at the corresponding position; feature maps (feature maps) are represented as a matrix of N L 'V, N being the number of kernels, L' ≦ L;
further, the converting the feature map (feature map) into a feature representation of local context dependency includes:
averaging the feature maps (feature maps) represented as a matrix of N L 'V to obtain a vector output of L' V;
converting the vector into a feature matrix with a specified size, namely a feature representation of local context correlation by utilizing a feedforward neural network layer;
the attention expression obtaining module is specifically configured to:
inputting the word vector into an improved multi-head attention model to calculate multi-head attention, wherein the improved multi-head attention model comprises local context features;
and splicing the multiple attention heads to obtain hidden layer state output as an attention representation.
4. A memory storing a plurality of instructions for implementing the method of any one of claims 1-2.
5. An electronic device comprising a processor and a memory coupled to the processor, the memory storing a plurality of instructions that are loadable and executable by the processor to enable the processor to perform the method according to any of claims 1-2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011351553.3A CN112487796B (en) | 2020-11-27 | 2020-11-27 | Method and device for sequence labeling and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011351553.3A CN112487796B (en) | 2020-11-27 | 2020-11-27 | Method and device for sequence labeling and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112487796A CN112487796A (en) | 2021-03-12 |
| CN112487796B true CN112487796B (en) | 2022-02-18 |
Family
ID=74935586
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011351553.3A Active CN112487796B (en) | 2020-11-27 | 2020-11-27 | Method and device for sequence labeling and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112487796B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112967112B (en) * | 2021-03-24 | 2022-04-29 | 武汉大学 | Electronic commerce recommendation method for self-attention mechanism and graph neural network |
| CN113901210B (en) * | 2021-09-15 | 2022-12-13 | 昆明理工大学 | Method for marking verbosity of Thai and Burma characters by using local multi-head attention to mechanism fused word-syllable pair |
Family Cites Families (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107220231A (en) * | 2016-03-22 | 2017-09-29 | 索尼公司 | Electronic equipment and method and training method for natural language processing |
| CN108717409A (en) * | 2018-05-16 | 2018-10-30 | 联动优势科技有限公司 | A kind of sequence labelling method and device |
| CN109635109B (en) * | 2018-11-28 | 2022-12-16 | 华南理工大学 | Sentence classification method based on LSTM and combined with part-of-speech and multi-attention mechanism |
| US11544259B2 (en) * | 2018-11-29 | 2023-01-03 | Koninklijke Philips N.V. | CRF-based span prediction for fine machine learning comprehension |
| GB201904167D0 (en) * | 2019-03-26 | 2019-05-08 | Benevolentai Tech Limited | Name entity recognition with deep learning |
| CN110083705B (en) * | 2019-05-06 | 2021-11-02 | 电子科技大学 | Multi-hop attention depth model, method, storage medium and terminal for target emotion classification |
| CN110472229B (en) * | 2019-07-11 | 2022-09-09 | 新华三大数据技术有限公司 | Sequence labeling model training method, electronic medical record processing method and related device |
| CN110674305B (en) * | 2019-10-10 | 2023-05-12 | 天津师范大学 | Commodity information classification method based on deep feature fusion model |
| CN111078895B (en) * | 2019-12-18 | 2023-04-18 | 江南大学 | Remote supervision entity relation extraction method based on denoising convolutional neural network |
| CN111783462B (en) * | 2020-06-30 | 2023-07-04 | 大连民族大学 | Chinese named entity recognition model and method based on double neural network fusion |
| CN112926323B (en) * | 2021-01-26 | 2024-02-02 | 江南大学 | Chinese named entity recognition method based on multistage residual convolution and attention mechanism |
-
2020
- 2020-11-27 CN CN202011351553.3A patent/CN112487796B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN112487796A (en) | 2021-03-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109887484B (en) | Dual learning-based voice recognition and voice synthesis method and device | |
| CN110134968B (en) | Poem generation method, device, equipment and storage medium based on deep learning | |
| CN111783462A (en) | Chinese named entity recognition model and method based on dual neural network fusion | |
| CN111062217B (en) | Language information processing method and device, storage medium and electronic equipment | |
| CN110866401A (en) | Chinese electronic medical record named entity identification method and system based on attention mechanism | |
| CN110795938B (en) | Text sequence word segmentation method, device and storage medium | |
| CN112487796B (en) | Method and device for sequence labeling and electronic equipment | |
| CN110968725B (en) | Image content description information generation method, electronic device and storage medium | |
| CN111611805B (en) | Auxiliary writing method, device, medium and equipment based on image | |
| CN114218945A (en) | Entity identification method, device, server and storage medium | |
| CN114218940B (en) | Text information processing and model training method, device, equipment and storage medium | |
| CN114445832A (en) | Character image recognition method and device based on global semantics and computer equipment | |
| CN111563380A (en) | Named entity identification method and device | |
| CN112183106A (en) | Semantic understanding method and device based on phoneme association and deep learning | |
| CN116739110A (en) | Large language model distillation method based on thinking chain | |
| JP2022145623A (en) | Method and device for presenting hint information and computer program | |
| CN111814479A (en) | Enterprise short form generation and model training method and device | |
| CN113870846A (en) | Speech recognition method, device and storage medium based on artificial intelligence | |
| CN112765985A (en) | Named entity identification method for specific field patent embodiment | |
| CN115687576B (en) | Keyword extraction method and device represented by theme constraint | |
| CN110210035B (en) | Sequence labeling method and device and training method of sequence labeling model | |
| CN114758330A (en) | Text recognition method and device, electronic equipment and storage medium | |
| CN112487811A (en) | Cascading information extraction system and method based on reinforcement learning | |
| CN114676684B (en) | Text error correction method and device, computer equipment and storage medium | |
| CN115358186B (en) | Generating method and device of slot label and storage medium |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20210324 Address after: B201d-1, 3rd floor, building 8, yard 1, Zhongguancun East Road, Haidian District, Beijing 100083 Applicant after: Beijing innovation Zhiyuan Technology Co.,Ltd. Address before: B201d-1, 3rd floor, building 8, yard 1, Zhongguancun East Road, Haidian District, Beijing 100083 Applicant before: Beijing Zhiyuan Artificial Intelligence Research Institute |
|
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20210621 Address after: 603a, 6th floor, building 6, yard 1, Zhongguancun East Road, Haidian District, Beijing 100083 Applicant after: Beijing Zhipu Huazhang Technology Co.,Ltd. Address before: B201d-1, 3rd floor, building 8, yard 1, Zhongguancun East Road, Haidian District, Beijing 100083 Applicant before: Beijing innovation Zhiyuan Technology Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CB03 | Change of inventor or designer information |
Inventor after: Meng Qian Inventor after: Liu Debing Inventor after: Chou Yu Inventor before: Meng Qian Inventor before: Tang Jie Inventor before: Liu Debing Inventor before: Chou Yu |
|
| CB03 | Change of inventor or designer information |