CN112735475B - Method and system for searching disease knowledge through voice - Google Patents
Method and system for searching disease knowledge through voice Download PDFInfo
- Publication number
- CN112735475B CN112735475B CN202011567638.5A CN202011567638A CN112735475B CN 112735475 B CN112735475 B CN 112735475B CN 202011567638 A CN202011567638 A CN 202011567638A CN 112735475 B CN112735475 B CN 112735475B
- Authority
- CN
- China
- Prior art keywords
- data
- disease
- semantic
- training
- name
- 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
- 201000010099 disease Diseases 0.000 title claims abstract description 418
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 title claims abstract description 418
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000012545 processing Methods 0.000 claims abstract description 58
- 238000007781 pre-processing Methods 0.000 claims abstract description 20
- 238000001914 filtration Methods 0.000 claims abstract description 14
- 230000008569 process Effects 0.000 claims abstract description 13
- 238000012549 training Methods 0.000 claims description 124
- 230000011218 segmentation Effects 0.000 claims description 31
- 208000024891 symptom Diseases 0.000 claims description 27
- 238000003745 diagnosis Methods 0.000 claims description 16
- 238000011282 treatment Methods 0.000 claims description 16
- 238000004422 calculation algorithm Methods 0.000 claims description 12
- 238000000605 extraction Methods 0.000 claims description 12
- 239000000284 extract Substances 0.000 claims description 4
- 230000009467 reduction Effects 0.000 claims description 4
- 230000000694 effects Effects 0.000 claims description 3
- 230000003044 adaptive effect Effects 0.000 claims description 2
- 238000013528 artificial neural network Methods 0.000 claims description 2
- 238000001514 detection method Methods 0.000 claims description 2
- 239000000203 mixture Substances 0.000 claims description 2
- 238000000926 separation method Methods 0.000 claims description 2
- 230000008030 elimination Effects 0.000 claims 1
- 238000003379 elimination reaction Methods 0.000 claims 1
- 230000007613 environmental effect Effects 0.000 claims 1
- 238000011277 treatment modality Methods 0.000 description 8
- 201000001245 periodontitis Diseases 0.000 description 7
- 230000006870 function Effects 0.000 description 6
- 201000004328 Pulpitis Diseases 0.000 description 5
- 206010037464 Pulpitis dental Diseases 0.000 description 5
- 208000002925 dental caries Diseases 0.000 description 5
- 208000007565 gingivitis Diseases 0.000 description 5
- 238000003058 natural language processing Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 208000002193 Pain Diseases 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 208000018035 Dental disease Diseases 0.000 description 2
- 206010018291 Gingival swelling Diseases 0.000 description 2
- 208000014151 Stomatognathic disease Diseases 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 230000008520 organization Effects 0.000 description 2
- 230000000241 respiratory effect Effects 0.000 description 2
- 230000008961 swelling Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 208000004371 toothache Diseases 0.000 description 2
- 208000020446 Cardiac disease Diseases 0.000 description 1
- 206010008469 Chest discomfort Diseases 0.000 description 1
- 208000032843 Hemorrhage Diseases 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 208000034158 bleeding Diseases 0.000 description 1
- 230000000740 bleeding effect Effects 0.000 description 1
- 230000000747 cardiac effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000002592 echocardiography Methods 0.000 description 1
- 230000004438 eyesight Effects 0.000 description 1
- 208000019622 heart disease Diseases 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 230000007170 pathology Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 230000029058 respiratory gaseous exchange Effects 0.000 description 1
- 210000002345 respiratory system Anatomy 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
- G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
- G10L25/54—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination for retrieval
-
- 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
- G06F16/38—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/381—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using identifiers, e.g. barcodes, RFIDs
-
- 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
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/18—Speech classification or search using natural language modelling
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/18—Speech classification or search using natural language modelling
- G10L15/1822—Parsing for meaning understanding
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Signal Processing (AREA)
- Library & Information Science (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The embodiment of the invention relates to a method and a system for searching disease knowledge through voice, wherein the method comprises the following steps: preprocessing the first voice data to generate first sentence audio data; performing first audio character recognition processing on the first sentence audio data to generate first sentence character data; performing first semantic tag identification processing on the first statement character data to generate a first semantic tag data set; performing first disease classification learning processing corresponding to the first label type data to generate a plurality of first disease name data and corresponding first disease probability data; generating a corresponding first disease knowledge data set according to each first disease name data; forming first search result data by each first disease name data, disease probability data and disease knowledge data set; and outputting the first search result data set. The embodiment of the invention saves unnecessary input process, saves information filtering time and improves user experience and information searching precision.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a method and a system for searching disease knowledge through voice.
Background
The old people pay more attention to various diseases and health information, related information is often searched, the current main searching mode is realized through a text input mode, and information filtering is needed to be carried out on massive searching results by individuals. The method is difficult for the old, and on one hand, due to the eyesight problem, the old has low typing speed and high error rate, and the searching effect is influenced; on the other hand, if too much information is fully filtered, the processing time is long, and the health of the old people is affected.
Disclosure of Invention
The invention aims to provide a method and a system for searching disease knowledge through voice, which are based on a preset disease knowledge base and are additionally provided with a voice recognition function and a disease classification learning model, so that an unnecessary input process is saved for a user, the time for filtering and screening information is saved for the user, and the user experience and the information searching precision are improved.
In order to achieve the above object, a first aspect of embodiments of the present invention provides a method for searching knowledge about diseases through voice, the method including:
the disease knowledge search system receives the first voice data, performs first voice preprocessing on the first voice data and generates first sentence audio data;
performing first audio word recognition processing on the first sentence audio data to generate first sentence word data;
performing first semantic tag identification processing on the first statement text data to generate a first semantic tag data set; the first set of semantic tag data comprises first tag type data and a plurality of first semantic tag data;
according to the multiple first semantic label data, performing first disease classification learning processing corresponding to the first label type data to generate multiple first disease name data and corresponding first disease probability data;
inquiring a first name and related information corresponding relation table reflecting the corresponding relation between the disease name and the related information of the disease according to each first disease name data to generate a corresponding first disease knowledge data set;
forming first search result data by each first disease name data, the first disease probability data corresponding to the first disease name data and the first disease knowledge data set;
and forming a first search result data set by all the first search result data and outputting the first search result data set.
Preferably, the disease knowledge search system receives the first voice data, performs first voice preprocessing on the first voice data, and generates first sentence audio data, which specifically includes:
and a data preprocessing module of the disease knowledge search system receives the first voice data, and performs first audio filtering and noise reduction processing on the first voice data to generate first sentence audio data.
Preferably, the performing a first audio word recognition process on the first sentence audio data to generate first sentence word data specifically includes:
and the voice recognition module of the disease knowledge search system inputs the first sentence audio data into a first acoustic language recognition model for recognition processing to generate the first sentence text data.
Preferably, the performing a first semantic tag identification process on the first sentence text data to generate a first semantic tag data set specifically includes:
the semantic recognition module of the disease knowledge search system inputs the first sentence text data into a first intelligent word segmentation recognition model for recognition processing to generate a plurality of first word segmentation data;
using the plurality of first participle data to query a first participle and semantic label corresponding relation table reflecting the corresponding relation between the participle and the semantic label to obtain a plurality of first semantic label data;
inquiring a first semantic label and label type corresponding relation table reflecting the corresponding relation of the semantic labels and the label types according to each first semantic label data to generate corresponding first inquiry label type data;
combining the first query tag type data with the same type into a type group in all the first query tag type data, and taking the tag type corresponding to the type group containing the first query tag type data with the largest quantity as the first tag type data;
composing the plurality of first semantic tag data from all of the first semantic tag data; and forming the first semantic tag data set by the first tag type data and the plurality of first semantic tag data.
Further, the querying, by using the plurality of first participle data, a first participle and semantic label correspondence table reflecting correspondence between participles and semantic labels to obtain a plurality of first semantic label data specifically includes:
polling all first participle and semantic label corresponding relation records in the first participle and semantic label corresponding relation table, and taking the currently polled first participle and semantic label corresponding relation record as a first current record; the first participle and semantic label corresponding relation table comprises a plurality of first participle and semantic label corresponding relation records; the first word segmentation and semantic label corresponding relation record comprises first word segmentation information and first semantic label information;
performing first matching processing with the first word segmentation information of the first current record by using the plurality of first word segmentation data; sequentially extracting first word segmentation data from the plurality of first word segmentation data to serve as first current word segmentation data; when the first current participle data is the same as the first participle information, the first matching processing is successful;
and when the first matching processing is successful, extracting the first semantic label information of the first current record to generate the first semantic label data.
Preferably, the performing, according to the plurality of first semantic tag data, first disease classification learning processing corresponding to the first tag type data to generate a plurality of first disease name data and corresponding first disease probability data specifically includes:
a disease learning module of the disease knowledge search system determines a corresponding first disease classification learning model according to the first label type data; inputting the plurality of first semantic label data into the first disease classification learning model for learning to obtain a plurality of groups of first learning output data groups; each set of the first learning output data includes the first disease name data and the corresponding first disease probability data.
Preferably, the first and second liquid crystal materials are,
the first name and related information corresponding relation table comprises a plurality of first name and related information corresponding relation records; the first name and related information corresponding relation record comprises first disease name information, first disease definition information, first disease symptom information, first disease cause information, first disease diagnosis mode information, first disease clinical expression information and first disease treatment mode information;
the first disease knowledge data set includes at least first disease definition data, first disease symptom data, first disease cause data, first disease diagnosis mode data, first disease clinical manifestation data, and first disease treatment mode data.
Preferably, the querying, according to each piece of the first disease name data, a first name and related information correspondence table that reflects correspondence between disease names and disease related information, and generating a corresponding first disease knowledge data set specifically include:
a disease knowledge extraction module of the disease knowledge search system polls all first name and related information corresponding relation records of the first name and related information corresponding relation table according to each first disease name data, and takes the currently polled first name and related information corresponding relation record as a second current record;
when each first disease name data is the same as the first disease name information of the second current record, extracting the first disease definition information as the corresponding first disease definition data, extracting the first disease symptom information as the corresponding first disease symptom data, extracting the first disease cause information as the corresponding first disease cause data, extracting the first disease diagnosis mode information as the corresponding first disease diagnosis mode data, extracting the first disease clinical expression information as the corresponding first disease clinical expression data, and extracting the first disease treatment mode information as the corresponding first disease treatment mode data from the second current record;
and the corresponding first disease knowledge data set is composed of the first disease definition data, the first disease symptom data, the first disease cause data, the first disease diagnosis mode data, the first disease clinical manifestation data and the first disease treatment mode data.
Preferably, before using the first disease classification learning model, the method further comprises:
the model training module of the disease knowledge search system performs learning model training processing on the first disease classification learning model by using semantic labels and an epidemic disease name training library; extracting multiple groups of semantic label training data corresponding to the specified epidemic disease name training data from the semantic label and epidemic disease name training library, and inputting the semantic label training data into the first disease classification learning model for training to obtain multiple groups of training output data groups; the semantic label and epidemic disease name training library comprises a plurality of semantic label training data and a plurality of epidemic disease name training data; each epidemic disease name training data corresponds to a plurality of semantic label training data; the training output data set comprises training output disease name data and training output disease probability data;
when in the multiunit training output data group, the probability is the highest training output disease probability data corresponds training output disease name data with appointed epidemic disease name training data is the same, and the probability is the highest training output disease probability data surpasses the training probability threshold value that sets for, and/or other when the correlation of training output disease name data and appointed epidemic disease name training data surpasses the training correlation threshold value that sets for, the training of learning model is handled successfully.
Preferably, after the disease knowledge search system outputs the first search result data set, the method further comprises:
a scoring processing module of the disease knowledge search system receives a first set of scoring data; the first set of scoring data comprises a plurality of first scoring data; the first set of scoring data corresponds to the first set of search result data; the first scoring data corresponds to the first search result data;
taking the plurality of first semantic label numbers as newly-added semantic label training data;
in the semantic label and epidemic disease name training library, taking the training disease name data corresponding to the first score data with the highest score as target training disease name data;
and adding the newly added semantic tag training data into the semantic tag and epidemic disease name training library, and establishing a corresponding relation between the newly added semantic tag training data and the target training disease name data.
A second aspect of an embodiment of the present invention provides a system for searching knowledge of a disease through speech, the system including:
the data preprocessing module is used for receiving the first voice data, performing first voice preprocessing on the first voice data and generating first statement audio data;
the voice recognition module is used for performing first audio character recognition processing on the first sentence audio data to generate first sentence character data;
the semantic identification module is used for carrying out first semantic tag identification processing on the first statement text data to generate a first semantic tag data set; the first set of semantic tag data comprises first tag type data and a plurality of first semantic tag data;
the disease learning module is used for performing first disease classification learning processing corresponding to the first label type data according to the plurality of first semantic label data to generate a plurality of first disease name data and corresponding first disease probability data;
the disease knowledge extraction module is used for inquiring a first name and related information corresponding relation table reflecting the corresponding relation between the disease name and the related information according to each first disease name data to generate a corresponding first disease knowledge data set;
the search result output module is used for combining each first disease name data, the corresponding first disease probability data and the corresponding first disease knowledge data set into first search result data; and forming a first search result data set by all the first search result data and outputting the first search result data set.
The embodiment of the invention provides a method and a system for searching disease knowledge through voice, which are based on a preset disease knowledge base and are additionally provided with a voice recognition function and a disease classification learning model, so that an unnecessary input process is saved for a user, the time for filtering and screening information is saved for the user, and the user experience and the information searching precision are improved.
Drawings
Fig. 1 is a schematic diagram of a method for searching knowledge of diseases by voice according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a system for searching knowledge of diseases through voice according to a second embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the present invention will be described in further detail with reference to the accompanying drawings, and it is apparent that the described embodiments are only a part of the embodiments of the present invention, not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
An embodiment of the present invention provides a method for searching knowledge of a disease through voice, and as shown in fig. 1, which is a schematic diagram of the method for searching knowledge of a disease through voice according to the embodiment of the present invention, the method mainly includes the following steps:
step 1, a disease knowledge search system receives first voice data, and performs first voice preprocessing on the first voice data to generate first sentence audio data;
the method specifically comprises the following steps: the data preprocessing module of the disease knowledge search system receives the first voice data, and carries out first audio filtering and noise reduction processing on the first voice data to generate first statement audio data.
Here, the disease knowledge search system may be understood as a system having speech semantic recognition and an intelligent knowledge base; the system comprises a data preprocessing module, a voice recognition module, a semantic recognition module, a disease learning module and a disease knowledge extraction module; the data preprocessing module is used for acquiring, denoising and filtering original voice data; the voice recognition module is used for carrying out voice recognition on the preprocessed audio data to obtain sentence and character data; the semantic recognition module carries out word segmentation and disease semantic recognition on the sentence character data, and all disease labels, namely semantic labels, and disease types with the maximum probability, namely label types are counted; the disease learning module is used for determining a disease classification learning model according to the label types, inputting all the counted disease labels into the disease classification learning model for deep learning, and finally obtaining a plurality of possible disease names and corresponding probabilities; the disease knowledge extraction module is used for extracting disease knowledge related to all possible diseases to serve as a final voice search result.
Here, in this step, the first Voice data is from a Voice recording device connected to the disease knowledge search system or a terminal device or a server storing original Voice data, and the data preprocessing module of the disease knowledge search system performs mute and noise separation processing on the Voice through a Voice Activity Detection algorithm (VAD); noise cancellation processing is performed on ambient noise, echoes, reverberation, and the like in the voice data using Least Mean Square (LMS) adaptive filtering, wiener filtering, and the like.
the method specifically comprises the following steps: and a voice recognition module of the disease knowledge search system inputs the first sentence audio data into a first acoustic language recognition model for recognition processing to generate first sentence text data.
Here, the first acoustic language recognition model used by the speech recognition module of the disease knowledge search system is commonly used as: 1) An acoustic Language recognition Model composed of a Hidden Markov Model (HMM) + Gaussian Mixture Model (GMM) + N-Gram Language Model/Chinese Language Model (CLM); 2) An acoustic language recognition model consisting of HMM + Deep Neural Network (DNN) + N-Gram/CLM; the first acoustic language recognition model extracts characteristic data of input first sentence audio data, performs pronunciation matching on the characteristic data to obtain a pronunciation data sequence with maximum probability, and performs language word and word recognition on the pronunciation data sequence to obtain a word string with maximum probability, namely first sentence character data.
wherein the first set of semantic tag data comprises first tag type data and a plurality of first semantic tag data;
here, the semantic recognition module of the disease knowledge search system extracts a tag type and a semantic tag related to a known disease from the first sentence text data;
the method specifically comprises the following steps: step 31, a semantic recognition module of the disease knowledge search system inputs the first sentence text data into a first intelligent participle recognition model for recognition processing to generate a plurality of first participle data;
here, the first intelligent word segmentation recognition model used by the semantic recognition module of the disease knowledge search system is an algorithm model based on Natural Language Processing (NLP), and commonly used are: a forward Maximum Matching (MM) algorithm model, a Reverse Maximum Matching (RMM) algorithm model, a Bi-directional Maximum Matching (Bi-directional Matching, BM) algorithm model, an HMM algorithm model, and a Conditional Random Field (CRF) algorithm model;
here, the NLP theory is a technical theory for processing, understanding, and using human language in the field of computer science and artificial intelligence, so as to achieve effective communication between a human and a computer; NLP can be basically divided into two parts: natural language decomposition processing and natural language generation processing; the embodiment of the invention mainly relates to a natural language decomposition processing part, in particular to a method for extracting participles from first original information by using a first artificial intelligent participle algorithm model based on an NLP theory; the word segmentation is the word of the minimum unit in a segment of text information, and the segment of text information comprises a plurality of word segmentations;
for example, the first sentence text data is "my toothache and swollen gum", the first sentence text data is segmented and refined by using the first intelligent segmentation recognition model, and nouns and verbs are used as refinement addition items in the refinement, and finally obtained first segmentation information is respectively: "my", "tooth", "pain", "get", "tooth", "bed", "swelling", "plus" toothache "," gum swelling ";
step 32, using the plurality of first participle data to inquire a first participle and semantic label corresponding relation table reflecting the corresponding relation between the participle and the semantic label to obtain a plurality of first semantic label data;
the first participle and semantic label corresponding relation table comprises a plurality of first participle and semantic label corresponding relation records; the first word segmentation and semantic label corresponding relation record comprises first word segmentation information and first semantic label information;
the method specifically comprises the following steps: polling the corresponding relation record of the first participle and the semantic label in the corresponding relation table of the first participle and the semantic label, and taking the corresponding relation record of the currently polled first participle and the semantic label as a first current record;
performing first matching processing on the first word segmentation information recorded at the first current time by using a plurality of first word segmentation data; sequentially extracting first word segmentation data from the plurality of first word segmentation data to serve as first current word segmentation data; when the first current participle data is the same as the first participle information, the first matching processing is successful; when the first matching processing is successful, extracting first semantic label information of a first current record to generate first semantic label data;
here, the corresponding relation table of the first participle and the semantic tag used by the semantic recognition module of the disease knowledge search system may be a database relation table or a data file; the words of the natural language are subjected to disease semantic labeling processing through the corresponding relation table of the first participle and the semantic labels, so that redundant data generated by repeated expression and approximate expression can be reduced; semantic tags here are actually tags related to disease symptoms, e.g., 195 for poor dental nerve perception, 196 for gum pathology, 197 for tooth bleeding symptoms, 279 for chest discomfort, 280 for poor breathing, etc.;
for example, the table of correspondence between the first participle and the semantic tag is as shown in table one, and the plurality of pieces of first participle information are respectively: "i", "tooth", "pain", "get", "tooth", "bed", "swelling", "plus" pain "," gum swelling "are given two first semantic label data: 195 and 196;
watch 1
Step 33, according to each first semantic tag data, inquiring a first semantic tag and tag type corresponding relation table reflecting the semantic tag and tag type corresponding relation, and generating corresponding first inquiry tag type data;
the first semantic label and label type corresponding relation table comprises a plurality of first semantic label and label type corresponding relation records; the first semantic label and label type corresponding relation record comprises second semantic label information and first label type information;
the method specifically comprises the following steps: polling a first semantic label and label type corresponding relation record of a first semantic label and label type corresponding relation table, and taking the currently polled first semantic label and label type corresponding relation record as a second current record;
when each first semantic label data is the same as second semantic label information of a second current record, extracting first label type information of the second current record as corresponding first query label type data;
here, the first semantic tag and tag type correspondence table used by the semantic identification module of the disease knowledge search system may be a database relationship table or a data file; querying the disease type corresponding to the disease semantic label through the first semantic label and label type corresponding relation table, wherein the disease type is actually a large class, for example, 11 represents dental related diseases, 21 represents heart related diseases, 31 represents respiratory related diseases, and the like;
for example, the table of correspondence between the first semantic tag and the tag type is shown in table two, and two first semantic tag data: 195 and 196, the two first query tag type data obtained are 11,11;
watch two
Step 34, merging the first query tag type data with the same type into a type group in all the first query tag type data, and taking the tag type corresponding to the type group containing the most amount of the first query tag type data as the first tag type data;
here, the disease category with the highest probability is selected from the disease categories;
for example, two first semantic tag data: 195 and 196; all corresponding first query tag type data are 11 and 11; generating a set of types including 11,11; if the tag type corresponding to the type group containing the largest amount of the first query tag type data is the tag type 11 of the type group, the first tag type data is 11;
step 35, forming a plurality of first semantic tag data by all the first semantic tag data; a first semantic tag data set is formed by first tag type data and a plurality of first semantic tag data.
Here, after steps 31-35 of step 3, the semantic recognition module of the disease knowledge search system performs further semantic analysis on the first sentence text data obtained in step 2, and the obtained first semantic tag data set includes the maximum probability disease category, that is, the first tag type data, and all semantic tags related to symptoms extracted from the original sentence.
Step 4, according to the plurality of first semantic tag data, performing first disease classification learning processing corresponding to the first tag type data to generate a plurality of first disease name data and corresponding first disease probability data;
the method specifically comprises the following steps: a disease learning module of the disease knowledge search system determines a corresponding first disease classification learning model according to the first label type data; inputting a plurality of first semantic label data into a first disease classification learning model for learning to obtain a plurality of groups of first learning output data groups; each set of first learning output data includes first disease name data and corresponding first disease probability data.
Here, the disease knowledge search system may have a plurality of disease classification learning models, such as a dental disease classification learning model for dental-related diseases, a cardiac disease classification learning model for cardiac-related diseases, a respiratory tract classification learning model for respiratory-related diseases, and the like; before each disease classification learning model is used, a model training module of a disease knowledge search system needs to train the disease classification learning model to be mature by using a semantic label and an epidemic disease name training library; the algorithm model adopted by the disease classification learning model is a random forest model commonly used, the type of input data can be classified and identified, and a plurality of possible classification results and the probability of each result are obtained; for example, the first tag type data is 11, a corresponding disease classification learning model, that is, a dental disease classification learning model, is selected, and for two first semantic tag data: 195 and 196, learning, the final calculation results are: periodontitis and its probable probability of 44%, gingivitis and its probable probability of 10.27%, pulpitis and its probable probability of 8.57%, and caries and its probable probability of 4.11%.
Step 5, according to each first disease name data, inquiring a first name and related information corresponding relation table reflecting the corresponding relation of the disease name and the related information of the disease, and generating a corresponding first disease knowledge data set;
the first name and related information corresponding relation table comprises a plurality of first name and related information corresponding relation records; the first name and related information corresponding relation record comprises first disease name information, first disease definition information, first disease symptom information, first disease cause information, first disease diagnosis mode information, first disease clinical expression information and first disease treatment mode information; the first disease knowledge data set at least comprises first disease definition data, first disease symptom data, first disease cause data, first disease diagnosis mode data, first disease clinical presentation data and first disease treatment mode data;
the method specifically comprises the following steps: a disease knowledge extraction module of the disease knowledge search system polls all first name and related information corresponding relation records of the first name and related information corresponding relation table according to each first disease name data, and takes the currently polled first name and related information corresponding relation record as a second current record;
when each first disease name data is the same as the first disease name information of the second current record, extracting first disease definition information from the second current record as corresponding first disease definition data, extracting first disease symptom information as corresponding first disease symptom data, extracting first disease cause information as corresponding first disease cause data, extracting first disease diagnosis mode information as corresponding first disease diagnosis mode data, extracting first disease clinical expression information as corresponding first disease clinical expression data, and extracting first disease treatment mode information as corresponding first disease treatment mode data;
and a corresponding first disease knowledge data set is composed of first disease definition data, first disease symptom data, first disease cause data, first disease diagnosis mode data, first disease clinical presentation data and first disease treatment mode data.
Here, the disease knowledge extraction module of the disease knowledge search system uses a first name and related information correspondence table which is actually a disease knowledge base, which may be a relational database, a form set composed of a plurality of database relationship tables, or a file set composed of a plurality of data files; in the first name and related information corresponding relation table, each first name and related information corresponding relation record records related information of a disease, including name, definition, common symptoms, etiology and inducement, diagnosis mode, clinical manifestation, treatment mode and the like; by taking the first disease name data as a query keyword, all relevant information can be extracted by querying the corresponding relation between the first name and the relevant information;
for example, 4 sets of first disease name data and corresponding first disease probability data are obtained from step 4: periodontitis and its probable probability of 44%, gingivitis and its probable probability of 10.27%, pulpitis and its probable probability of 8.57%, caries and its probable probability of 4.11%; then 4 first disease knowledge data sets are obtained by step 5: a disease knowledge data set (including definitions, common symptoms, etiologies and causes, diagnostic modalities, clinical manifestations, treatment modalities, etc.) about periodontitis, a disease knowledge data set (including definitions, common symptoms, etiologies and causes, diagnostic modalities, clinical manifestations, treatment modalities, etc.) about gingivitis, a disease knowledge data set (including definitions, common symptoms, etiologies and causes, diagnostic modalities, clinical manifestations, treatment modalities, etc.) about pulpitis, a disease knowledge data set (including definitions, common symptoms, etiologies and causes, diagnostic modalities, clinical manifestations, treatment modalities, etc.) about dental caries.
And 6, forming first search result data by each first disease name data, the corresponding first disease probability data and the corresponding first disease knowledge data set.
For example, the disease knowledge extraction module of the disease knowledge search system obtains 4 first search result data from 4 sets of first disease name data and corresponding first disease probability data, and 4 first disease knowledge data sets:
1 st first search result data: periodontitis, likely probability 44%, a knowledge data set of diseases about periodontitis (including definitions, common symptoms, causes and causes, diagnostic modalities, clinical manifestations, treatment modalities, etc.);
2 nd first search result data: gingivitis, probability of being 10.27%, a set of knowledge data about the disease of gingivitis (including definitions, common symptoms, etiologies and causes, diagnostic modalities, clinical manifestations, treatment modalities, etc.);
3 rd first search result data: pulpitis, likely probability of 8.57%, a set of disease knowledge data about pulpitis (including definitions, common symptoms, etiologies and causes, diagnostic modalities, clinical manifestations, treatment modalities, etc.);
4 th first search result data: caries, probability 4.11%, disease knowledge data set about caries (including definitions, common symptoms, etiologies and causes, diagnostic modalities, clinical manifestations, treatment modalities, etc.).
And 7, forming a first search result data set by all the first search result data and outputting the first search result data set.
Here, the disease knowledge extraction module of the disease knowledge search system assembles all the obtained first search result data into a first search result data set to feed back to the user.
In addition, the disease knowledge search system further comprises a model training module, and in the embodiment of the invention, before each disease classification learning model is put into use, the model training module needs to use the semantic label and the epidemic disease name training library to train each disease classification learning model, wherein the training process is briefly described as follows:
a1, a model training module of a disease knowledge search system extracts multiple groups of semantic tag training data corresponding to specified epidemic disease name training data from a semantic tag and epidemic disease name training library, inputs the semantic tag training data into a first disease classification learning model and trains the semantic tag training data to obtain multiple groups of training output data;
the semantic label and epidemic disease name training library comprises a plurality of semantic label training data and a plurality of epidemic disease name training data; each epidemic disease name training data corresponds to a plurality of semantic label training data; the training output data set includes training output disease name data and training output disease probability data.
The training data in the semantic label and the epidemic disease name training database are verified data, wherein the corresponding relation between the semantic label training data and the epidemic disease name training data is verified to be correct; the data of the semantic label and the epidemic disease name training library can be effective test data provided by a third-party testing organization, and can also be medical data acquired from a medical organization; the larger the training data amount is, the more accurate the corresponding relation is, and the higher the precision of the trained model is.
And step A2, when training output disease name data corresponding to training output disease probability data with the highest probability in the multiple groups of training output data groups are the same as training data of the specified epidemic disease name, and the training output disease probability data with the highest probability exceeds a set training probability threshold, and/or the correlation between other training output disease name data and the training data of the specified epidemic disease name exceeds a set training correlation threshold, successfully training the learning model.
Here are the conditions described for terminating training during model training: on the premise of ensuring that the designated disease name data appears and the probability is maximum, the precision of the probability is high enough to exceed a set training probability threshold value as a reference; and the relevance between other classification results and the main classification result can be considered, and the higher the relevance is, the higher the calculation precision of the model is, and the relevance exceeds a set training relevance threshold value as a reference.
In addition, the disease knowledge search system further includes a score processing module, and after the first search result data set is output, the score processing module automatically enriches the semantic label and the epidemic disease name training library according to the score of the user on the output result, which is specifically described as follows:
b1, a grading processing module of the disease knowledge search system receives a first grading data set;
wherein the first set of scoring data comprises a plurality of first scoring data; the first set of scoring data corresponds to the first set of search result data; the first score data corresponds to the first search result data.
For example, after the disease knowledge search system displays 4 pieces of first search result data to the user, the disease knowledge search system also provides the user with an evaluation function, and the evaluation is ranked in three levels: best, generally, not; if the score of the user on the 1 st first search result data is the most consistent, the score on the 2 nd is generally the consistent, and the scores on the 3 rd and 4 th are not consistent, the score processing module may obtain 4 first score data in the first score data set as: best, general, not.
Step B2, taking a plurality of first semantic label numbers as newly-added semantic label training data; in the semantic label and epidemic disease name training library, training disease name data corresponding to the first scoring data with the highest score are used as target training disease name data; and adding new semantic label training data into the semantic label and epidemic disease name training library, and establishing a corresponding relation between the new semantic label training data and the target training disease name data.
For example, if the disease name data corresponding to the 1 st first search result data with the highest score, that is, the score that is the most matched with the disease name data, is "periodontitis", from the 4 first score data in the first score data set, this step will convert the current two first semantic label numbers obtained by the user's voice into: 195 and 196, which are added to the semantic tag and epidemic name training library and associated with the training disease name data, specifically "periodontitis", in the library, which is actually to add effective training data to the semantic tag and epidemic name training library.
A second embodiment of the present invention provides a system for searching knowledge of diseases by voice, which is used to implement the system functions of the disease knowledge searching system in the foregoing embodiments, and specifically, as shown in fig. 2, which is a schematic structural diagram of a system for searching knowledge of diseases by voice according to the second embodiment of the present invention, the system 20 mainly includes: the system comprises a data preprocessing module 201, a voice recognition module 202, a semantic recognition module 203, a disease learning module 204, a disease knowledge extraction module 205 and a search result output module 206.
The data preprocessing module 201 is configured to receive first voice data, perform first voice preprocessing on the first voice data, and generate first sentence audio data.
The voice recognition module 202 is configured to perform a first audio character recognition process on the first sentence audio data to generate first sentence character data.
The semantic identification module 203 is configured to perform a first semantic tag identification process on the first statement text data to generate a first semantic tag data set; the first set of semantic tag data includes a first tag type data and a plurality of first semantic tag data.
The disease learning module 204 is configured to perform a first disease classification learning process corresponding to the first tag type data according to the plurality of first semantic tag data, and generate a plurality of first disease name data and corresponding first disease probability data.
The disease knowledge extraction module 205 is configured to query, according to each first disease name data, a first name and related information correspondence table that reflects a correspondence between a disease name and disease related information, and generate a corresponding first disease knowledge data set.
The search result output module 206 is configured to assemble each first disease name data, and the first disease probability data and the first disease knowledge data set corresponding to the first disease name data into first search result data; and a first search result data set is formed by all the first search result data and is output.
Here, in the system for searching knowledge of disease through voice provided in the second embodiment of the present invention, the functions of the modules are the same as those of the modules corresponding to the system for searching knowledge of disease in the first embodiment, which is not further described herein.
The embodiment of the invention provides a method and a system for searching disease knowledge through voice, which are based on a preset disease knowledge base and are additionally provided with a voice recognition function and a disease classification learning model, so that an unnecessary input process is saved for a user, the time for filtering and screening information is saved for the user, and the user experience and the information searching precision are improved.
Those of skill would further appreciate that the various illustrative components and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied in hardware, a software module executed by a processor, or a combination of the two. A software module may reside in Random Access Memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
The above-mentioned embodiments, objects, technical solutions and advantages of the present invention are further described in detail, it should be understood that the above-mentioned embodiments are only examples of the present invention, and are not intended to limit the scope of the present invention, and any modifications, equivalent substitutions, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (8)
1. A method for searching knowledge of diseases by speech, the method comprising:
the disease knowledge searching system receives the first voice data, performs first voice preprocessing on the first voice data and generates first sentence audio data;
performing first audio word recognition processing on the first sentence audio data to generate first sentence word data;
performing first semantic tag identification processing on the first statement text data to generate a first semantic tag data set; the first set of semantic tag data comprises first tag type data and a plurality of first semantic tag data;
according to the multiple first semantic label data, performing first disease classification learning processing corresponding to the first label type data to generate multiple first disease name data and corresponding first disease probability data;
according to each first disease name data, inquiring a first name and related information corresponding relation table reflecting the corresponding relation between the disease name and the related information of the disease to generate a corresponding first disease knowledge data set;
forming first search result data by each first disease name data, the first disease probability data corresponding to the first disease name data and the first disease knowledge data set;
forming a first search result data set by all the first search result data and outputting the first search result data set;
the disease knowledge search system receives first voice data, performs first voice preprocessing on the first voice data, and generates first sentence audio data, and the method specifically includes:
a data preprocessing module of the disease knowledge search system receives the first voice data, and performs first audio filtering and noise reduction processing on the first voice data to generate first statement audio data;
the performing a first audio word recognition process on the first sentence audio data to generate first sentence word data specifically includes:
the voice recognition module of the disease knowledge search system inputs the first sentence audio data into a first acoustic language recognition model for recognition processing to generate first sentence text data;
when the data preprocessing module carries out first audio filtering and noise reduction processing on the first voice data, carrying out mute and noise separation processing on voice through a voice activity detection algorithm; noise elimination processing is carried out on environmental noise, echo, reverberation and the like in the voice data by using least mean square adaptive filtering and wiener filtering;
when the voice recognition module inputs first sentence audio data into a first acoustic language recognition model for recognition processing, the first acoustic language recognition model extracts characteristic data of the input first sentence audio data, pronounces and matches the characteristic data to obtain a pronunciation data sequence with maximum probability, and then performs language word and word recognition on the pronunciation data sequence to obtain a word string with maximum probability as corresponding first sentence text data; the first acoustic language recognition model comprises an acoustic language recognition model consisting of a hidden Markov model, a Gaussian mixture model, an N-Gram language model or a Chinese language model, and an acoustic language recognition model consisting of the hidden Markov model, a deep neural network, the N-Gram language model or the Chinese language model.
2. The method for searching knowledge about diseases through voice according to claim 1, wherein the performing a first semantic tag recognition process on the first sentence text data to generate a first semantic tag data set specifically includes:
the semantic recognition module of the disease knowledge search system inputs the first sentence character data into a first intelligent word segmentation recognition model for recognition processing to generate a plurality of first word segmentation data;
using the plurality of first participle data to inquire a first participle and semantic label corresponding relation table reflecting the corresponding relation between the participle and the semantic label to obtain a plurality of first semantic label data;
inquiring a first semantic label and label type corresponding relation table reflecting the corresponding relation of the semantic labels and the label types according to each first semantic label data to generate corresponding first inquiry label type data;
combining the first query tag type data with the same type into a type group in all the first query tag type data, and taking the tag type corresponding to the type group containing the first query tag type data with the largest quantity as the first tag type data;
composing the plurality of first semantic tag data from all of the first semantic tag data; and forming the first semantic tag data set by the first tag type data and the plurality of first semantic tag data.
3. The method of claim 2, wherein the searching a table of correspondence between the first participles and the semantic tags, which reflects the correspondence between the participles and the semantic tags, using the plurality of first participles to obtain a plurality of first semantic tag data, comprises:
polling all first participle and semantic label corresponding relation records in the first participle and semantic label corresponding relation table, and taking the currently polled first participle and semantic label corresponding relation record as a first current record; the first participle and semantic label corresponding relation table comprises a plurality of first participle and semantic label corresponding relation records; the first word segmentation and semantic label corresponding relation record comprises first word segmentation information and first semantic label information;
performing first matching processing with the first segmentation information of the first current record by using the plurality of first segmentation data; sequentially extracting first word segmentation data from the plurality of first word segmentation data to serve as first current word segmentation data; when the first current participle data is the same as the first participle information, the first matching processing is successful;
and when the first matching processing is successful, extracting the first semantic label information of the first current record to generate the first semantic label data.
4. The method for searching knowledge about diseases through voice according to claim 1, wherein the performing a first disease classification learning process corresponding to the first tag type data according to the plurality of first semantic tag data to generate a plurality of first disease name data and corresponding first disease probability data specifically includes:
a disease learning module of the disease knowledge search system determines a corresponding first disease classification learning model according to the first label type data; inputting the plurality of first semantic label data into the first disease classification learning model for learning to obtain a plurality of groups of first learning output data groups; each set of the first learning output data includes the first disease name data and the corresponding first disease probability data.
5. The method for searching knowledge of illness by voice according to claim 1,
the first name and related information corresponding relation table comprises a plurality of first name and related information corresponding relation records; the first name and related information corresponding relation record comprises first disease name information, first disease definition information, first disease symptom information, first disease cause information, first disease diagnosis mode information, first disease clinical expression information and first disease treatment mode information;
the first disease knowledge data set at least comprises first disease definition data, first disease symptom data, first disease cause data, first disease diagnosis mode data, first disease clinical presentation data and first disease treatment mode data;
the querying, according to each of the first disease name data, a first name and related information correspondence table that reflects a correspondence between a disease name and disease related information, and generating a corresponding first disease knowledge data set specifically include:
a disease knowledge extraction module of the disease knowledge search system polls all first name and related information corresponding relation records of the first name and related information corresponding relation table according to each first disease name data, and takes the currently polled first name and related information corresponding relation record as a second current record;
when each first disease name data is the same as the first disease name information of the second current record, extracting the first disease definition information as the corresponding first disease definition data, extracting the first disease symptom information as the corresponding first disease symptom data, extracting the first disease cause information as the corresponding first disease cause data, extracting the first disease diagnosis mode information as the corresponding first disease diagnosis mode data, extracting the first disease clinical expression information as the corresponding first disease clinical expression data, and extracting the first disease treatment mode information as the corresponding first disease treatment mode data from the second current record;
and the corresponding first disease knowledge data set is composed of the first disease definition data, the first disease symptom data, the first disease cause data, the first disease diagnosis mode data, the first disease clinical presentation data and the first disease treatment mode data.
6. The method for searching knowledge of illness through speech of claim 4, wherein prior to using the first illness classification learning model, the method further comprises:
the model training module of the disease knowledge search system performs learning model training processing on the first disease classification learning model by using semantic labels and an epidemic disease name training library; extracting a plurality of groups of semantic label training data corresponding to the specified epidemic disease name training data from the semantic labels and the epidemic disease name training library, and inputting the semantic label training data into the first disease classification learning model for training to obtain a plurality of groups of training output data sets; the semantic label and epidemic disease name training library comprises a plurality of semantic label training data and a plurality of epidemic disease name training data; each epidemic disease name training data corresponds to a plurality of semantic label training data; the training output data set comprises training output disease name data and training output disease probability data;
when the training output disease name data which is the highest in probability in the multiple groups of training output data sets and corresponds to the training output disease probability data is the same as the training data of the appointed epidemic disease name and is the highest in probability the training output disease probability data exceeds a set training probability threshold value and/or other degrees of correlation between the training output disease name data and the training data of the appointed epidemic disease name exceeds a set training degree of correlation threshold value, the training of the learning model is successful.
7. The method for searching knowledge of illness through speech of claim 6, wherein after the illness knowledge search system outputs the first search result data set, the method further comprises:
a scoring processing module of the disease knowledge search system receives a first set of scoring data; the first set of scoring data comprises a plurality of first scoring data; the first set of scoring data corresponds to the first set of search result data; the first scoring data corresponds to the first search result data;
taking the plurality of first semantic label numbers as newly-added semantic label training data;
training disease name data corresponding to the first grading data with the highest grading is used as target training disease name data in the semantic label and epidemic disease name training library;
and adding the newly added semantic tag training data into the semantic tag and epidemic disease name training library, and establishing a corresponding relation between the newly added semantic tag training data and the target training disease name data.
8. A system for implementing the method for searching knowledge about diseases by voice according to any one of claims 1 to 7, the system comprising:
the data preprocessing module is used for receiving the first voice data and performing first voice preprocessing on the first voice data to generate first sentence audio data;
the voice recognition module is used for performing first audio character recognition processing on the first sentence audio data to generate first sentence character data;
the semantic identification module is used for carrying out first semantic tag identification processing on the first statement character data to generate a first semantic tag data set; the first set of semantic tag data comprises first tag type data and a plurality of first semantic tag data;
the disease learning module is used for performing first disease classification learning processing corresponding to the first label type data according to the plurality of first semantic label data to generate a plurality of first disease name data and corresponding first disease probability data;
the disease knowledge extraction module is used for inquiring a first name and related information corresponding relation table reflecting the corresponding relation between the disease name and the related information according to each first disease name data to generate a corresponding first disease knowledge data set;
the search result output module is used for combining each first disease name data, the corresponding first disease probability data and the corresponding first disease knowledge data set into first search result data; and forming a first search result data set by all the first search result data and outputting the first search result data set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011567638.5A CN112735475B (en) | 2020-12-25 | 2020-12-25 | Method and system for searching disease knowledge through voice |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011567638.5A CN112735475B (en) | 2020-12-25 | 2020-12-25 | Method and system for searching disease knowledge through voice |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112735475A CN112735475A (en) | 2021-04-30 |
| CN112735475B true CN112735475B (en) | 2023-02-21 |
Family
ID=75616691
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011567638.5A Active CN112735475B (en) | 2020-12-25 | 2020-12-25 | Method and system for searching disease knowledge through voice |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112735475B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101404035A (en) * | 2008-11-21 | 2009-04-08 | 北京得意音通技术有限责任公司 | Information search method based on text or voice |
| CN105760399A (en) * | 2014-12-19 | 2016-07-13 | 华为软件技术有限公司 | Data retrieval method and device |
| CN106682411A (en) * | 2016-12-22 | 2017-05-17 | 浙江大学 | Method for converting physical examination diagnostic data into disease label |
| CN108052659A (en) * | 2017-12-28 | 2018-05-18 | 北京百度网讯科技有限公司 | Searching method, device and electronic equipment based on artificial intelligence |
| CN108182262A (en) * | 2018-01-04 | 2018-06-19 | 华侨大学 | Intelligent Answer System construction method and system based on deep learning and knowledge mapping |
| CN111274373A (en) * | 2020-01-16 | 2020-06-12 | 山东大学 | Electronic medical record question-answering method and system based on knowledge graph |
| CN111522910A (en) * | 2020-04-14 | 2020-08-11 | 浙江大学 | Intelligent semantic retrieval method based on cultural relic knowledge graph |
| CN111597308A (en) * | 2020-05-19 | 2020-08-28 | 中国电子科技集团公司第二十八研究所 | Knowledge graph-based voice question-answering system and application method thereof |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030105638A1 (en) * | 2001-11-27 | 2003-06-05 | Taira Rick K. | Method and system for creating computer-understandable structured medical data from natural language reports |
| CN106557653B (en) * | 2016-11-15 | 2017-09-22 | 合肥工业大学 | A kind of portable medical intelligent medical guide system and method |
| US11132361B2 (en) * | 2018-11-20 | 2021-09-28 | International Business Machines Corporation | System for responding to complex user input queries using a natural language interface to database |
-
2020
- 2020-12-25 CN CN202011567638.5A patent/CN112735475B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101404035A (en) * | 2008-11-21 | 2009-04-08 | 北京得意音通技术有限责任公司 | Information search method based on text or voice |
| CN105760399A (en) * | 2014-12-19 | 2016-07-13 | 华为软件技术有限公司 | Data retrieval method and device |
| CN106682411A (en) * | 2016-12-22 | 2017-05-17 | 浙江大学 | Method for converting physical examination diagnostic data into disease label |
| CN108052659A (en) * | 2017-12-28 | 2018-05-18 | 北京百度网讯科技有限公司 | Searching method, device and electronic equipment based on artificial intelligence |
| CN108182262A (en) * | 2018-01-04 | 2018-06-19 | 华侨大学 | Intelligent Answer System construction method and system based on deep learning and knowledge mapping |
| CN111274373A (en) * | 2020-01-16 | 2020-06-12 | 山东大学 | Electronic medical record question-answering method and system based on knowledge graph |
| CN111522910A (en) * | 2020-04-14 | 2020-08-11 | 浙江大学 | Intelligent semantic retrieval method based on cultural relic knowledge graph |
| CN111597308A (en) * | 2020-05-19 | 2020-08-28 | 中国电子科技集团公司第二十八研究所 | Knowledge graph-based voice question-answering system and application method thereof |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112735475A (en) | 2021-04-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111461176B (en) | Multi-mode fusion method, device, medium and equipment based on normalized mutual information | |
| CN117253576B (en) | Outpatient electronic medical record generation method based on Chinese medical large model | |
| Lipping et al. | Crowdsourcing a dataset of audio captions | |
| Syed et al. | Automated recognition of alzheimer’s dementia using bag-of-deep-features and model ensembling | |
| Levitan et al. | Combining Acoustic-Prosodic, Lexical, and Phonotactic Features for Automatic Deception Detection. | |
| CN111833853A (en) | Voice processing method and device, electronic equipment and computer readable storage medium | |
| Gontijo et al. | Grapheme—phoneme probabilities in British English | |
| CN111180025B (en) | Method, device and inquiry system for representing text vectors of medical records | |
| CN110675292A (en) | Child language ability evaluation method based on artificial intelligence | |
| CN115148210A (en) | Voice recognition system and voice recognition method | |
| Tasnim et al. | Depac: a corpus for depression and anxiety detection from speech | |
| Baese-Berk et al. | Intelligibility as a measure of speech perception: Current approaches, challenges, and recommendations | |
| CN114927126A (en) | Scheme output method, device and equipment based on semantic analysis and storage medium | |
| Ren et al. | Evaluation of the pain level from speech: Introducing a novel pain database and benchmarks | |
| CN112735475B (en) | Method and system for searching disease knowledge through voice | |
| CN116959754A (en) | Feature extraction method of structured interview recording transcribed text based on intention slots | |
| Nikadon et al. | BERTAgent: The development of a novel tool to quantify agency in textual data | |
| Brown | Y-ACCDIST: An automatic accent recognition system for forensic applications | |
| Biswas et al. | Can ChatGPT be Your Personal Medical Assistant? | |
| Shirali-Shahreza et al. | Better replacement for TTS naturalness evaluation | |
| CN113593523B (en) | Speech detection method and device based on artificial intelligence and electronic equipment | |
| CN112735412B (en) | Method and system for searching information according to voice instruction | |
| CN113761899A (en) | Medical text generation method, device, equipment and storage medium | |
| CN110033778B (en) | Real-time identification and correction system for lie state | |
| Syed et al. | Static vs. dynamic modelling of acoustic speech features for detection of dementia |
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 |