CN113744831A - Online medical application purchasing system - Google Patents
Online medical application purchasing system Download PDFInfo
- Publication number
- CN113744831A CN113744831A CN202110962675.4A CN202110962675A CN113744831A CN 113744831 A CN113744831 A CN 113744831A CN 202110962675 A CN202110962675 A CN 202110962675A CN 113744831 A CN113744831 A CN 113744831A
- Authority
- CN
- China
- Prior art keywords
- subunit
- convolution
- medical application
- application software
- input
- 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.)
- Pending
Links
- 238000000605 extraction Methods 0.000 claims description 51
- 238000011176 pooling Methods 0.000 claims description 38
- 238000005070 sampling Methods 0.000 claims description 17
- 238000006243 chemical reaction Methods 0.000 claims description 16
- 238000010586 diagram Methods 0.000 claims description 13
- 230000004913 activation Effects 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 12
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 6
- 230000011218 segmentation Effects 0.000 claims description 6
- 230000002776 aggregation Effects 0.000 claims description 5
- 238000004220 aggregation Methods 0.000 claims description 5
- 239000012634 fragment Substances 0.000 claims description 3
- 108010001267 Protein Subunits Proteins 0.000 description 10
- 238000000034 method Methods 0.000 description 6
- 230000009286 beneficial effect Effects 0.000 description 5
- 239000000284 extract Substances 0.000 description 4
- 230000000717 retained effect Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- 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
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
-
- 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
- G10L15/02—Feature extraction for speech recognition; Selection of recognition unit
-
- 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
- G10L15/16—Speech classification or search using artificial neural networks
-
- 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
- G10L15/26—Speech to text systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- General Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Multimedia (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Epidemiology (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Probability & Statistics with Applications (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention discloses an online medical application purchasing system, which comprises: the system comprises a user login module, a user retrieval module, a medical application software recommendation module, a medical application software database and a display module; the invention designs 3 input modes, the voice, the picture and the characters meet the user requirements of various conditions, simultaneously, the application software required by the user is accurately and quickly searched out through the matching of the keywords, the matching rate is high, and the user can conveniently and quickly purchase the medical application software.
Description
Technical Field
The invention relates to the field of medical application purchasing systems, in particular to an online medical application purchasing system.
Background
The hospital has a large demand for purchasing medical application software, but the conventional medical application purchasing system requires that a purchaser spends a large amount of time and energy for screening, and has the problems of single input method and long retrieval time, so that the time and cost for purchasing a user are high, and the software required by the user cannot be accurately screened.
Disclosure of Invention
Aiming at the defects in the prior art, the online medical application purchasing system provided by the invention solves the problems of single input method, long retrieval time and inaccurate retrieval of the conventional medical purchasing system.
In order to achieve the purpose of the invention, the invention adopts the technical scheme that: an online medical application purchasing system, comprising: the system comprises a user login module, a user retrieval module, a medical application software recommendation module, a medical application software database and a display module;
the user login module is used for user login; the medical application software recommending module collects application software names of medical application software purchased by a user, further analyzes the requirements of the user to obtain recommended medical application software, and recommends the user; the user retrieval module is used for matching application software names of the medical application software in the medical application software database according to the input of the user after the login user views the medical application software recommended by the medical application software recommendation module, so that all medical application software meeting the input of the user are obtained; the display module sequentially presents all medical application software meeting the user input.
Further, the user input includes: voice input, image input, and text input; the user retrieval module comprises: a voice-to-character model, an image-to-character model, a keyword extraction model and a matching model;
the voice-to-character model is used for converting voice input by a user into characters; the image-to-character model is used for converting an image input by a user into characters; the keyword extraction model is used for extracting keywords from characters input by a user or characters converted from a voice to character model or characters converted from an image to character model to obtain a keyword set; the matching model is used for matching each noun in the keyword set with the application software name of the medical application software in the medical application software database to obtain all medical application software meeting the user input.
Further, the speech-to-text model comprises: the system comprises a voice segmentation unit, a voice feature extraction unit, a voice feature conversion unit and a text conversion unit;
the voice segmentation unit is used for segmenting voice to obtain multi-frame sound segments, and the multi-frame sound segments are overlapped between adjacent frames; the voice feature extraction unit is used for converting each frame of voice fragment into a multi-dimensional vector to obtain voice features; the voice feature conversion unit is used for converting the voice features into a phoneme set consisting of an initial consonant and a final vowel; the text conversion unit is used for converting the phoneme set into characters.
Further, the voice feature conversion unit comprises a first LSTM subunit, a second LSTM subunit, a first fully-connected subunit, a softmax classification subunit and a CTC loss function subunit which are connected in sequence;
the loss function L of the CTC loss function subunit is as follows:
wherein N is the frame number of the sound segment, M is the category number of the initial and final, yijWhether the class j belongs to the real class of the input i-th frame sound clip, pijThe probability that the sound clip belongs to the category j is the ith frame.
Further, the image-to-text model includes: the device comprises a line scanning feature extraction unit, a column scanning feature extraction unit and a BP neural network mapping unit; the line scanning feature extraction unit is used for extracting features of line pixels of the image to obtain line features; the column scanning feature extraction unit is used for extracting features of column pixels of the image to obtain column features; and the BP neural network mapping unit is used for analyzing the row characteristics and the column characteristics to obtain corresponding characters.
The beneficial effects of the above further scheme are: the image-to-character model designed by the invention starts from the row pixels and the column pixels of the image respectively, and then extracts the characteristics of two aspects, so that the characteristic extraction of the image is more comprehensive.
Further, the line scanning feature extraction unit and the column scanning feature extraction unit each include: the system comprises an input subunit, a first convolution subunit, a second convolution subunit, a third convolution subunit, a fourth convolution subunit, a fifth convolution subunit, a sixth convolution subunit and a second full-connection subunit;
the output end of the input subunit is respectively connected with the input end of the first convolution subunit, the output end of the first convolution subunit, the input end of the second convolution subunit, the output end of the second convolution subunit, the input end of the third convolution subunit, the output end of the third convolution subunit, the input end of the fourth convolution subunit, the output end of the fourth convolution subunit, the input end of the fifth convolution subunit, the output end of the fifth convolution subunit and the input end of the sixth convolution subunit; the output end of the sixth convolution subunit is connected with the input end of the second full-connection subunit; the input end of the input subunit is used as the input end of a line scanning feature extraction unit or a column scanning feature extraction unit; the output end of the second full-connection subunit is used as the output end of the line scanning feature extraction unit or the column scanning feature extraction unit;
the first convolution subunit, the second convolution subunit, the third convolution subunit, the fourth convolution subunit, the fifth convolution subunit and the sixth convolution subunit have the same structure and all comprise: the device comprises a convolution layer, a down-sampling layer, a projection layer, an activation layer, a first pooling layer, a second pooling layer and an addition layer;
the input end of the convolution layer is used as the input end of the first convolution subunit, the second convolution subunit, the third convolution subunit, the fourth convolution subunit, the fifth convolution subunit or the sixth convolution subunit, and the output end of the convolution layer is connected with the input end of the down-sampling layer; the output end of the down-sampling layer is connected with the input end of the projection layer; the output end of the projection layer is connected with the input end of the activation layer; the output end of the activation layer is respectively connected with the input end of the first pooling layer and the input end of the second pooling layer; the output end of the first pooling layer is connected with the first input end of the addition layer; the output end of the second pooling layer is connected with the second input end of the addition layer; and the output end of the addition layer is used as the output end of the first convolution subunit, the second convolution subunit, the third convolution subunit, the fourth convolution subunit, the fifth convolution subunit or the sixth convolution subunit.
The beneficial effects of the above further scheme are: by adding the image data output by the input subunit into each convolution subunit for processing, the image data output by the input subunit contains richest information, and key information is prevented from being lost in the processes of down-sampling and pooling.
Further, the convolution layer is used for carrying out convolution processing on the image data to obtain characteristic map data; the down-sampling layer is used for down-sampling the characteristic diagram data to obtain the thumbnail characteristic diagram data; the projection layer is used for projecting the thumbnail feature map data to a high-dimensional space to obtain a high-dimensional projection map; the activation layer is used for carrying out nonlinear processing on the high-dimensional projection drawing to obtain an intermediate characteristic drawing; the first pooling layer is used for carrying out weighted average operation on the intermediate characteristic diagram to obtain a first intermediate characteristic vector; the second pooling layer is used for carrying out global significance aggregation weighting on the intermediate feature map to obtain a second intermediate feature vector formed by combining and weighting the most significant features in the intermediate feature map; and the addition layer is used for splicing the first intermediate characteristic vector and the second intermediate characteristic vector to obtain an output characteristic vector.
The beneficial effects of the above further scheme are: the processing modes of the first pooling layer and the second pooling layer for the intermediate feature maps are different, the first pooling layer performs weighted average processing on the intermediate feature maps to reduce the number of parameters, the second pooling layer performs global significance aggregation weighting on the intermediate feature maps to retain the most significant features, and through the combination of the two pooling modes, the removal of the significant features in the pooling process is avoided, and the features are retained in the maximum range.
Further, the keyword extraction model is used for removing connecting words in the characters to obtain a plurality of nouns; the nouns are combined into a keyword set; the matching model is used for respectively matching a plurality of nouns in the keyword set with each application software name of the medical application software in the medical application software database to obtain a matching rate; and the display module sequentially presents all medical application software meeting the input of the user according to the matching rate.
The beneficial effects of the above further scheme are: the keyword extraction model is used for removing connection words in the text, such as: a connection small software for printing removes useless words such as a connection word 'one used for, small software' and the like, only extracts nouns 'printing' and 'connection', avoids the interference of the useless words, simultaneously matches the 'printing' and the 'connection' with application software names, if the two words can be matched, the matching rate is high, if only one word can be matched, the matching rate is low, if neither word can be matched, no related word exists in a medical application software database, namely, no related application software exists; in this way, the application software which is most needed by the user is quickly screened out.
In conclusion, the beneficial effects of the invention are as follows: the invention designs 3 input modes, the voice, the picture and the characters meet the user requirements of various conditions, simultaneously, the application software required by the user is accurately and quickly searched out through the matching of the keywords, the matching rate is high, and the user can conveniently and quickly purchase the medical application software.
Drawings
FIG. 1 is a system block diagram of an online medical application purchasing system;
FIG. 2 is a schematic diagram of a structure of a speech feature conversion unit;
FIG. 3 is a schematic structural diagram of a line scanning feature extraction unit and a column scanning feature extraction unit;
fig. 4 is a schematic diagram of the structure of the convolution subunit.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined in the appended claims, and all matters produced by the invention using the inventive concept are protected.
As shown in fig. 1, an online medical application purchasing system includes: the system comprises a user login module, a user retrieval module, a medical application software recommendation module, a medical application software database and a display module;
the user login module is used for user login; the medical application software recommending module collects application software names of medical application software purchased by a user, further analyzes the requirements of the user to obtain recommended medical application software, and recommends the user; the user retrieval module is used for matching application software names of the medical application software in the medical application software database according to the input of the user after the login user views the medical application software recommended by the medical application software recommendation module, so that all medical application software meeting the input of the user are obtained; the display module sequentially presents all medical application software meeting the user input.
The user's inputs include: voice input, image input, and text input; the user retrieval module comprises: a voice-to-character model, an image-to-character model, a keyword extraction model and a matching model;
the voice-to-character model is used for converting voice input by a user into characters; the image-to-character model is used for converting an image input by a user into characters; the keyword extraction model is used for extracting keywords from characters input by a user or characters converted from a voice to character model or characters converted from an image to character model to obtain a keyword set; the matching model is used for matching each noun in the keyword set with the application software name of the medical application software in the medical application software database to obtain all medical application software meeting the user input.
The speech-to-text model comprises: the system comprises a voice segmentation unit, a voice feature extraction unit, a voice feature conversion unit and a text conversion unit;
the voice segmentation unit is used for segmenting voice to obtain multi-frame sound segments, and the multi-frame sound segments are overlapped between adjacent frames; the voice feature extraction unit is used for converting each frame of voice fragment into a multi-dimensional vector to obtain voice features; the voice feature conversion unit is used for converting the voice features into a phoneme set consisting of an initial consonant and a final vowel; the text conversion unit is used for converting the phoneme set into characters.
As shown in fig. 2, the speech feature conversion unit includes a first LSTM subunit, a second LSTM subunit, a first fully-connected subunit, a softmax classification subunit, and a CTC loss function subunit, which are connected in sequence;
the loss function L of the CTC loss function subunit is as follows:
wherein N is the frame number of the sound segment, M is the category number of the initial and final, yijWhether the class j belongs to the real class of the input i-th frame sound clip, pijThe probability that the sound clip belongs to the category j is the ith frame.
The image-to-text model comprises: the device comprises a line scanning feature extraction unit, a column scanning feature extraction unit and a BP neural network mapping unit; the line scanning feature extraction unit is used for extracting features of line pixels of the image to obtain line features; the column scanning feature extraction unit is used for extracting features of column pixels of the image to obtain column features; and the BP neural network mapping unit is used for analyzing the row characteristics and the column characteristics to obtain corresponding characters.
The image-to-character model designed by the invention starts from the row pixels and the column pixels of the image respectively, and then extracts the characteristics of two aspects, so that the characteristic extraction of the image is more comprehensive.
As shown in fig. 3, each of the line scan feature extraction unit and the column scan feature extraction unit includes: the system comprises an input subunit, a first convolution subunit, a second convolution subunit, a third convolution subunit, a fourth convolution subunit, a fifth convolution subunit, a sixth convolution subunit and a second full-connection subunit;
the output end of the input subunit is respectively connected with the input end of the first convolution subunit, the output end of the first convolution subunit, the input end of the second convolution subunit, the output end of the second convolution subunit, the input end of the third convolution subunit, the output end of the third convolution subunit, the input end of the fourth convolution subunit, the output end of the fourth convolution subunit, the input end of the fifth convolution subunit, the output end of the fifth convolution subunit and the input end of the sixth convolution subunit; the output end of the sixth convolution subunit is connected with the input end of the second full-connection subunit; the input end of the input subunit is used as the input end of a line scanning feature extraction unit or a column scanning feature extraction unit; the output end of the second full-connection subunit is used as the output end of the line scanning feature extraction unit or the column scanning feature extraction unit;
as shown in fig. 4, the first convolution sub-unit, the second convolution sub-unit, the third convolution sub-unit, the fourth convolution sub-unit, the fifth convolution sub-unit, and the sixth convolution sub-unit have the same structure, and each of the first convolution sub-unit, the second convolution sub-unit, the third convolution sub-unit, the fourth convolution sub-unit, the fifth convolution sub-unit, and the sixth convolution sub-unit includes: the device comprises a convolution layer, a down-sampling layer, a projection layer, an activation layer, a first pooling layer, a second pooling layer and an addition layer;
the input end of the convolution layer is used as the input end of the first convolution subunit, the second convolution subunit, the third convolution subunit, the fourth convolution subunit, the fifth convolution subunit or the sixth convolution subunit, and the output end of the convolution layer is connected with the input end of the down-sampling layer; the output end of the down-sampling layer is connected with the input end of the projection layer; the output end of the projection layer is connected with the input end of the activation layer; the output end of the activation layer is respectively connected with the input end of the first pooling layer and the input end of the second pooling layer; the output end of the first pooling layer is connected with the first input end of the addition layer; the output end of the second pooling layer is connected with the second input end of the addition layer; and the output end of the addition layer is used as the output end of the first convolution subunit, the second convolution subunit, the third convolution subunit, the fourth convolution subunit, the fifth convolution subunit or the sixth convolution subunit.
By adding the image data output by the input subunit into each convolution subunit for processing, the image data output by the input subunit contains richest information, and key information is prevented from being lost in the processes of down-sampling and pooling.
The convolution layer is used for carrying out convolution processing on the image data to obtain characteristic map data; the down-sampling layer is used for down-sampling the characteristic diagram data to obtain the thumbnail characteristic diagram data; the projection layer is used for projecting the thumbnail feature map data to a high-dimensional space to obtain a high-dimensional projection map; the activation layer is used for carrying out nonlinear processing on the high-dimensional projection drawing to obtain an intermediate characteristic drawing; the first pooling layer is used for carrying out weighted average operation on the intermediate characteristic diagram to obtain a first intermediate characteristic vector; the second pooling layer is used for carrying out global significance aggregation weighting on the intermediate feature map to obtain a second intermediate feature vector formed by combining and weighting the most significant features in the intermediate feature map; and the addition layer is used for splicing the first intermediate characteristic vector and the second intermediate characteristic vector to obtain an output characteristic vector.
The processing modes of the first pooling layer and the second pooling layer for the intermediate feature maps are different, the first pooling layer performs weighted average processing on the intermediate feature maps to reduce the number of parameters, the second pooling layer performs global significance aggregation weighting on the intermediate feature maps to retain the most significant features, and through the combination of the two pooling modes, the removal of the significant features in the pooling process is avoided, and the features are retained in the maximum range.
The keyword extraction model is used for removing connecting words in the characters to obtain a plurality of nouns; the nouns are combined into a keyword set; the matching model is used for respectively matching a plurality of nouns in the keyword set with each application software name of the medical application software in the medical application software database to obtain a matching rate; and the display module sequentially presents all medical application software meeting the input of the user according to the matching rate.
The keyword extraction model is used for removing connection words in the text, such as: a connection small software for printing removes useless words such as a connection word 'one used for, small software' and the like, only extracts nouns 'printing' and 'connection', avoids the interference of the useless words, simultaneously matches the 'printing' and the 'connection' with application software names, if the two words can be matched, the matching rate is high, if only one word can be matched, the matching rate is low, if neither word can be matched, no related word exists in a medical application software database, namely, no related application software exists; in this way, the application software which is most needed by the user is quickly screened out.
Claims (8)
1. An online medical application purchasing system, comprising: the system comprises a user login module, a user retrieval module, a medical application software recommendation module, a medical application software database and a display module;
the user login module is used for user login; the medical application software recommending module collects application software names of medical application software purchased by a user, further analyzes the requirements of the user to obtain recommended medical application software, and recommends the user; the user retrieval module is used for matching application software names of the medical application software in the medical application software database according to the input of the user after the login user views the medical application software recommended by the medical application software recommendation module, so that all medical application software meeting the input of the user are obtained; the display module sequentially presents all medical application software meeting the user input.
2. The online medical application purchasing system of claim 1, wherein the user's input includes: voice input, image input, and text input; the user retrieval module comprises: a voice-to-character model, an image-to-character model, a keyword extraction model and a matching model;
the voice-to-character model is used for converting voice input by a user into characters; the image-to-character model is used for converting an image input by a user into characters; the keyword extraction model is used for extracting keywords from characters input by a user or characters converted from a voice to character model or characters converted from an image to character model to obtain a keyword set; the matching model is used for matching each noun in the keyword set with the application software name of the medical application software in the medical application software database to obtain all medical application software meeting the user input.
3. The online medical application purchasing system of claim 2, wherein the speech-to-text model comprises: the system comprises a voice segmentation unit, a voice feature extraction unit, a voice feature conversion unit and a text conversion unit;
the voice segmentation unit is used for segmenting voice to obtain multi-frame sound segments, and the multi-frame sound segments are overlapped between adjacent frames; the voice feature extraction unit is used for converting each frame of voice fragment into a multi-dimensional vector to obtain voice features; the voice feature conversion unit is used for converting the voice features into a phoneme set consisting of an initial consonant and a final vowel; the text conversion unit is used for converting the phoneme set into characters.
4. The online medical application purchasing system of claim 3, wherein the speech feature conversion unit includes a first LSTM subunit, a second LSTM subunit, a first fully connected subunit, a softmax classification subunit and a CTC loss function subunit connected in sequence;
the loss function L of the CTC loss function subunit is as follows:
wherein N is the frame number of the sound segment, M is the category number of the initial and final, yijWhether the class j belongs to the real class of the input i-th frame sound clip, pijThe probability that the sound clip belongs to the category j is the ith frame.
5. The online medical application purchasing system of claim 2, wherein the image-to-text model includes: the device comprises a line scanning feature extraction unit, a column scanning feature extraction unit and a BP neural network mapping unit; the line scanning feature extraction unit is used for extracting features of line pixels of the image to obtain line features; the column scanning feature extraction unit is used for extracting features of column pixels of the image to obtain column features; and the BP neural network mapping unit is used for analyzing the row characteristics and the column characteristics to obtain corresponding characters.
6. The online medical application purchasing system of claim 5, wherein the line scan feature extraction unit and the column scan feature extraction unit each include: the system comprises an input subunit, a first convolution subunit, a second convolution subunit, a third convolution subunit, a fourth convolution subunit, a fifth convolution subunit, a sixth convolution subunit and a second full-connection subunit;
the output end of the input subunit is respectively connected with the input end of the first convolution subunit, the output end of the first convolution subunit, the input end of the second convolution subunit, the output end of the second convolution subunit, the input end of the third convolution subunit, the output end of the third convolution subunit, the input end of the fourth convolution subunit, the output end of the fourth convolution subunit, the input end of the fifth convolution subunit, the output end of the fifth convolution subunit and the input end of the sixth convolution subunit; the output end of the sixth convolution subunit is connected with the input end of the second full-connection subunit; the input end of the input subunit is used as the input end of a line scanning feature extraction unit or a column scanning feature extraction unit; the output end of the second full-connection subunit is used as the output end of the line scanning feature extraction unit or the column scanning feature extraction unit;
the first convolution subunit, the second convolution subunit, the third convolution subunit, the fourth convolution subunit, the fifth convolution subunit and the sixth convolution subunit have the same structure and all comprise: the device comprises a convolution layer, a down-sampling layer, a projection layer, an activation layer, a first pooling layer, a second pooling layer and an addition layer;
the input end of the convolution layer is used as the input end of the first convolution subunit, the second convolution subunit, the third convolution subunit, the fourth convolution subunit, the fifth convolution subunit or the sixth convolution subunit, and the output end of the convolution layer is connected with the input end of the down-sampling layer; the output end of the down-sampling layer is connected with the input end of the projection layer; the output end of the projection layer is connected with the input end of the activation layer; the output end of the activation layer is respectively connected with the input end of the first pooling layer and the input end of the second pooling layer; the output end of the first pooling layer is connected with the first input end of the addition layer; the output end of the second pooling layer is connected with the second input end of the addition layer; and the output end of the addition layer is used as the output end of the first convolution subunit, the second convolution subunit, the third convolution subunit, the fourth convolution subunit, the fifth convolution subunit or the sixth convolution subunit.
7. The online medical application purchasing system of claim 6, wherein the convolution layer is configured to perform convolution processing on the image data to obtain feature map data; the down-sampling layer is used for down-sampling the characteristic diagram data to obtain the thumbnail characteristic diagram data; the projection layer is used for projecting the thumbnail feature map data to a high-dimensional space to obtain a high-dimensional projection map; the activation layer is used for carrying out nonlinear processing on the high-dimensional projection drawing to obtain an intermediate characteristic drawing; the first pooling layer is used for carrying out weighted average operation on the intermediate characteristic diagram to obtain a first intermediate characteristic vector; the second pooling layer is used for carrying out global significance aggregation weighting on the intermediate feature map to obtain a second intermediate feature vector formed by combining and weighting the most significant features in the intermediate feature map; and the addition layer is used for splicing the first intermediate characteristic vector and the second intermediate characteristic vector to obtain an output characteristic vector.
8. The online medical application purchasing system of claim 2, wherein the keyword extraction model is configured to remove conjunctions from the text to obtain a plurality of nouns; the nouns are combined into a keyword set; the matching model is used for respectively matching a plurality of nouns in the keyword set with each application software name of the medical application software in the medical application software database to obtain a matching rate; and the display module sequentially presents all medical application software meeting the input of the user according to the matching rate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110962675.4A CN113744831A (en) | 2021-08-20 | 2021-08-20 | Online medical application purchasing system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110962675.4A CN113744831A (en) | 2021-08-20 | 2021-08-20 | Online medical application purchasing system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113744831A true CN113744831A (en) | 2021-12-03 |
Family
ID=78732036
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110962675.4A Pending CN113744831A (en) | 2021-08-20 | 2021-08-20 | Online medical application purchasing system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113744831A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101281534A (en) * | 2008-05-28 | 2008-10-08 | 叶睿智 | Method for searching multimedia resource based on audio content retrieval |
| JP2015114710A (en) * | 2013-12-09 | 2015-06-22 | 株式会社東芝 | Medical information system and method for searching recommended application |
| CN107066477A (en) * | 2016-12-13 | 2017-08-18 | 合网络技术（北京）有限公司 | A kind of method and device of intelligent recommendation video |
| CN109977258A (en) * | 2019-02-21 | 2019-07-05 | 中国科学院西安光学精密机械研究所 | Cross-module state searching classification device model, searching system and the search method of image and voice |
| CN110110116A (en) * | 2019-04-02 | 2019-08-09 | 浙江工业大学 | A kind of trademark image retrieval method for integrating depth convolutional network and semantic analysis |
-
2021
- 2021-08-20 CN CN202110962675.4A patent/CN113744831A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101281534A (en) * | 2008-05-28 | 2008-10-08 | 叶睿智 | Method for searching multimedia resource based on audio content retrieval |
| JP2015114710A (en) * | 2013-12-09 | 2015-06-22 | 株式会社東芝 | Medical information system and method for searching recommended application |
| CN107066477A (en) * | 2016-12-13 | 2017-08-18 | 合网络技术（北京）有限公司 | A kind of method and device of intelligent recommendation video |
| CN109977258A (en) * | 2019-02-21 | 2019-07-05 | 中国科学院西安光学精密机械研究所 | Cross-module state searching classification device model, searching system and the search method of image and voice |
| CN110110116A (en) * | 2019-04-02 | 2019-08-09 | 浙江工业大学 | A kind of trademark image retrieval method for integrating depth convolutional network and semantic analysis |
Non-Patent Citations (2)
| Title |
|---|
| 姚舜才等: "《TensorFlow深度学习从入门到进阶》", 31 March 2020, 西安电子科技大学出版社, pages: 108 - 281 * |
| 杨博雄等: "《深度学习理论与实践》", 30 September 2020, 北京邮电大学出版社, pages: 71 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112818861B (en) | Emotion classification method and system based on multi-mode context semantic features | |
| US10755048B2 (en) | Artificial intelligence based method and apparatus for segmenting sentence | |
| CN111700608B (en) | Electrocardiosignal multi-classification method and device | |
| CN113641820A (en) | Visual angle level text emotion classification method and system based on graph convolution neural network | |
| CN111738169B (en) | Handwriting formula recognition method based on end-to-end network model | |
| CN112347284B (en) | Combined trademark image retrieval method | |
| CN109299227B (en) | Information query method and device based on voice recognition | |
| CN111694940A (en) | User report generation method and terminal equipment | |
| CN113094578A (en) | Deep learning-based content recommendation method, device, equipment and storage medium | |
| CN112836702B (en) | Text recognition method based on multi-scale feature extraction | |
| CN111125457A (en) | Deep cross-modal Hash retrieval method and device | |
| CN111459977A (en) | Conversion of natural language queries | |
| CN117076693A (en) | Method for constructing digital human teacher multi-mode large language model pre-training discipline corpus | |
| CN113627190A (en) | Visualized data conversion method and device, computer equipment and storage medium | |
| CN115481679A (en) | Multi-modal emotion analysis method and system | |
| CN111259223B (en) | News recommendation and text classification method based on emotion analysis model | |
| US20210271705A1 (en) | Generating accurate and natural captions for figures | |
| CN117271558A (en) | Language query model construction method, query language acquisition method and related devices | |
| CN112069402A (en) | Personalized comment recommendation method based on emotion and graph convolution neural network | |
| CN113744831A (en) | Online medical application purchasing system | |
| CN114064873A (en) | Method and device for building FAQ knowledge base in insurance field and electronic equipment | |
| CN112002306B (en) | Speech class recognition method and device, electronic equipment and readable storage medium | |
| CN114116953A (en) | Efficient semantic expansion retrieval method and device based on word vectors and storage medium | |
| CN111312215B (en) | Natural voice emotion recognition method based on convolutional neural network and binaural characterization | |
| CN110648666B (en) | Method and system for improving conference transcription performance based on conference outline |
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 |