CN108806671B - Semantic analysis, device and electronic equipment - Google Patents

Semantic analysis, device and electronic equipment Download PDF

Info

Publication number
CN108806671B
CN108806671B CN201810534587.2A CN201810534587A CN108806671B CN 108806671 B CN108806671 B CN 108806671B CN 201810534587 A CN201810534587 A CN 201810534587A CN 108806671 B CN108806671 B CN 108806671B
Authority
CN
China
Prior art keywords
text information
semantic
information
phrase
voice messaging
Prior art date
Application number
CN201810534587.2A
Other languages
Chinese (zh)
Other versions
CN108806671A (en
Inventor
李成君
仇志雄
应旭河
Original Assignee
杭州认识科技有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 杭州认识科技有限公司 filed Critical 杭州认识科技有限公司
Priority to CN201810534587.2A priority Critical patent/CN108806671B/en
Publication of CN108806671A publication Critical patent/CN108806671A/en
Application granted granted Critical
Publication of CN108806671B publication Critical patent/CN108806671B/en

Links

Classifications

    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00Speech recognition
    • G10L15/08Speech classification or search
    • G10L15/18Speech classification or search using natural language modelling
    • G10L15/1822Parsing for meaning understanding
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00Speech recognition
    • G10L15/08Speech classification or search
    • G10L15/18Speech classification or search using natural language modelling
    • G10L15/183Speech classification or search using natural language modelling using context dependencies, e.g. language models
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00Speech recognition
    • G10L15/26Speech to text systems
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L25/00Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
    • G10L25/48Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
    • G10L25/51Speech 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

Abstract

The present invention provides a kind of semantic analysis, device and electronic equipments, are related to semantic analysis technology field, which includes: the first voice messaging for receiving user's input;Text information is converted by the first voice messaging;Text information is handled using verbal model;The semantic understanding result that semantic analysis generates text information is carried out to treated text information using the deep learning model of building;The semantic understanding result of text information is exported.Therefore, technical solution provided by the invention, the technical issues of traditional semantic analysis existing in the prior art cannot systematically carry out complicated semantic analysis can be alleviated, propose a kind of semantic analysis of systematization, the accuracy that the semantic analysis of complicated voice input can be improved, promotes the development of human-computer interaction intelligent.

Description

Semantic analysis, device and electronic equipment

Technical field

The present invention relates to semantic analysis technology fields, more particularly, to a kind of semantic analysis, device and electronic equipment.

Background technique

With the development of science and technology, intellectual technology is rapidly developed and is popularized in electronic field.Speech recognition technology is intelligence How an important ring for energy technology, accurately identify that the voice of user is the development trend of intellectual technology.Currently, speech recognition skill The development of art largely improves the level of human-computer interaction, and semantic analysis technology is as the crucial portion for understanding natural language Point, the important task for how sufficiently analyzing and understanding that the input of user's natural language is semantic is carry, therefore for the intelligence of intelligence system Change degree has conclusive effect.However, traditional semantic analysis effect in terms of the semantic analysis that simple voice inputs Or it is good, but do not have the semantic analysis ability of systematization for complicated voice input, it is not able to satisfy and increasingly improves Human-computer interaction intelligent demand, the accuracy for how improving the semantic analysis of complicated voice input becomes human-computer interaction urgently The technical issues of solution.

Summary of the invention

In view of this, the purpose of the present invention is to provide semantic analysis, device and electronic equipment, to alleviate existing skill Traditional semantic analysis present in art cannot systematically carry out the technical issues of complicated semantic analysis.

In a first aspect, the embodiment of the invention provides a kind of semantic analysis, comprising:

Receive the first voice messaging of user's input;

Text information is converted by first voice messaging;

The text information is handled using verbal model;

Semantic analysis is carried out to treated text information using the deep learning model of building and generates text information Semantic understanding result;

The semantic understanding result of text information is exported.

With reference to first aspect, the embodiment of the invention provides the first possible embodiments of first aspect, wherein institute It states and converts text information for first voice messaging, specifically include:

Judge whether first voice messaging is received pronunciation information;

If it is not, first voice messaging is then converted to received pronunciation information;

Text information is converted by the received pronunciation information.

With reference to first aspect, the embodiment of the invention provides second of possible embodiments of first aspect, wherein institute It states and the text information is handled using verbal model, specifically include:

Character segmentation is carried out to text information using verbal model, filtering, classification, part of speech analysis, part-of-speech tagging, extracts and marks Label, obtain multiple participle phrases.

With reference to first aspect, the embodiment of the invention provides the third possible embodiments of first aspect, wherein institute It states and the semantic reason that semantic analysis generates text information is carried out to treated text information using the deep learning model of building Solution is as a result, specifically include:

Context understanding is carried out to treated text information using the deep learning models coupling application scenarios of building It is disambiguated with semanteme, generates the semantic understanding result of text information.

The third possible embodiment with reference to first aspect, the embodiment of the invention provides the 4th kind of first aspect Possible embodiment, wherein the deep learning models coupling application scenarios using building believe treated text Breath carries out context understanding and semantic disambiguation, generates the semantic understanding of text information as a result, specifically including:

Multiple phrase combination contexts of treated text information are carried out up and down using the deep learning model of building Unity and coherence in writing solution, semantic disambiguation;Obtain the semantic results of multiple phrases;

The semantic results of multiple phrases are compared with the phrase of knowledge mapping respectively, obtain the similarity of each phrase Value, using the highest phrase of similarity value as the semantic results of each phrase, obtains the semantic results of multiple phrases;

The semantic results of multiple phrases are combined, the semantic understanding result of text information is generated.

The third possible embodiment with reference to first aspect, the embodiment of the invention provides the 5th kind of first aspect Possible embodiment, wherein the deep learning models coupling application scenarios using building believe treated text Breath carries out context understanding and semantic disambiguation, generates the semantic understanding of text information as a result, specifically including:

Multiple phrase combination contexts of treated text information are carried out up and down using the deep learning model of building Unity and coherence in writing solution, semantic disambiguation;Obtain the semantic results of multiple phrases;

By the semantic results combination knowledge mapping of multiple phrases, the internal relation and/or logical relation of multiple phrases are analyzed, Generate the semantic understanding result of text information.

With reference to first aspect, the embodiment of the invention provides the 6th kind of possible embodiments of first aspect, wherein institute It states and exports the semantic understanding result of text information, specifically include:

The semantic understanding result of text information is exported in the form of text;

And/or the semantic understanding result of text information is exported in the form of received pronunciation;

And/or the semantic understanding result of text information is exported in the form of picture;

And/or the semantic understanding result of text information is exported in the form of video;

And/or the semantic understanding result of text information is exported in the form of hyperlink.

With reference to first aspect, the embodiment of the invention provides the 7th kind of possible embodiments of first aspect, wherein also Include:

In the training process and/or application process of the deep learning model, aid mark is carried out by manual intervention, To improve the understanding accuracy of participle.

Second aspect, the embodiment of the present invention also provide a kind of semantic analysis device, comprising:

Receiving module, for receiving the first voice messaging of user's input;

Conversion module, for converting text information for first voice messaging;

Processing module, for being handled using verbal model the text information;

It is raw to carry out semantic analysis to treated text information for the deep learning model using building for analysis module At the semantic understanding result of text information;

Output module, for exporting the semantic understanding result of text information.

The third aspect the embodiment of the invention also provides a kind of electronic equipment, including memory, processor and is stored in institute The computer program that can be run on memory and on the processor is stated, the processor executes real when the computer program The step of semantic analysis described in existing any one of above-mentioned first aspect and its possible embodiment.

Fourth aspect, the embodiment of the invention provides a kind of meters of non-volatile program code that can be performed with processor Calculation machine readable medium, said program code make the processor execute the aforementioned semantic analysis referred to.

The embodiment of the present invention bring it is following the utility model has the advantages that

In semantic analysis provided in an embodiment of the present invention, device and electronic equipment, wherein the semantic analysis packet It includes: receiving the first voice messaging of user's input;Text information is converted by the first voice messaging;Using verbal model to text Information is handled;Semantic analysis is carried out to treated text information using the deep learning model of building and generates text letter The semantic understanding result of breath;The semantic understanding result of text information is exported.Therefore, technology provided in an embodiment of the present invention Scheme, can alleviate traditional semantic analysis existing in the prior art cannot systematically carry out complicated semantic analysis Technical problem proposes a kind of semantic analysis scheme of systematization, and the standard of the semantic analysis of complicated voice input can be improved True property promotes the development of human-computer interaction intelligent.

Other features and advantages of the present invention will illustrate in the following description, also, partly become from specification It obtains it is clear that understand through the implementation of the invention.The objectives and other advantages of the invention are in specification, claims And specifically noted structure is achieved and obtained in attached drawing.

To enable the above objects, features and advantages of the present invention to be clearer and more comprehensible, preferred embodiment is cited below particularly, and cooperate Appended attached drawing, is described in detail below.

Detailed description of the invention

It, below will be to specific in order to illustrate more clearly of the specific embodiment of the invention or technical solution in the prior art Embodiment or attached drawing needed to be used in the description of the prior art be briefly described, it should be apparent that, it is described below Attached drawing is some embodiments of the present invention, for those of ordinary skill in the art, before not making the creative labor It puts, is also possible to obtain other drawings based on these drawings.

Fig. 1 is a kind of flow chart of semantic analysis provided in an embodiment of the present invention;

Fig. 2 is the flow chart of another semantic analysis provided in an embodiment of the present invention;

Fig. 3 is that the embodiment of the invention provides a kind of schematic diagrames of semantic analysis device;

Fig. 4 is the schematic diagram of a kind of electronic equipment provided in an embodiment of the present invention.

Specific embodiment

In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with attached drawing to the present invention Technical solution be clearly and completely described, it is clear that described embodiments are some of the embodiments of the present invention, rather than Whole embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not making creative work premise Under every other embodiment obtained, shall fall within the protection scope of the present invention.

Currently, the development of speech recognition technology largely improves the level of human-computer interaction, and semantic analysis technology is made For the key component for understanding natural language, the important task for how sufficiently analyzing and understanding that the input of user's natural language is semantic is carry, Therefore there is conclusive effect for the intelligence degree of intelligence system.However, traditional semantic analysis is in simple voice Effect or good in terms of the semantic analysis of input, but do not have the semantic analysis of systematization for complicated voice input Ability is not able to satisfy increasing human-computer interaction intelligent demand, how to improve the semantic analysis of complicated voice input Accuracy becomes human-computer interaction technical problem urgently to be resolved, is based on this, a kind of semantic analysis side provided in an embodiment of the present invention Method, device and electronic equipment, can alleviate traditional semantic analysis existing in the prior art cannot systematically carry out The technical issues of complicated semantic analysis, can be improved the accuracy of the semantic analysis of complicated voice input.

For convenient for understanding the present embodiment, first to a kind of semantic analysis disclosed in the embodiment of the present invention into Row is discussed in detail.

Embodiment one:

As shown in Figure 1, being specifically included the embodiment of the invention provides a kind of semantic analysis:

Step S101: the first voice messaging of user's input is received;

Here the first voice messaging can be received pronunciation information, such as mandarin, be also possible to non-standard language message Breath, such as dialect or foreign language;It can also be the voice messaging including non-default language, such as be mingled with foreign language or dialect Mandarin etc..

Step S102: text information is converted by the first voice messaging;

Specifically, step S102 is mainly realized by following steps:

1) judge whether first voice messaging is received pronunciation information;

Specifically, judging whether first voice messaging includes non-default language;The non-default language includes dialect Or foreign language;

2) if it is not, first voice messaging is then converted to received pronunciation information;

The step 2) is main to be realized by following steps:

A is not if it is determined that first voice messaging is for received pronunciation information;Determine that first voice messaging includes When non-default language, first voice messaging is extracted using dialect model, is divided, it is non-to obtain at least one Preset language phrase;And record the location information that non-default language phrase is located in the first voice messaging;

Here dialect model includes dialect model or foreign language model etc.;Non-default language phrase includes dialecticism Group or foreign language phrase;

Specifically, if it is determined that first voice messaging is not to utilize dialect model or foreign language then for received pronunciation information Model etc. respectively extracts first voice messaging, divides, and obtains at least one dialect phrase or foreign language phrase;And Record dialect phrase or foreign language phrase are located at the location information in the first voice messaging;

B searches phrase and corresponds to table and disambiguate in conjunction with context progress semanteme corresponding with the non-default language phrase to obtain Standard language phrase;

Step B specifically includes the following steps:

B1 is according to position of the non-default language phrase in the voice messaging to judge the non-default language word The part of speech of group;

B2 is based on the part of speech and searches phrase corresponding to table to obtain the meaning of a word corresponding with the non-default language phrase;

B3 judges whether the meaning of a word is multiple;

Here multiple refer to two or more.

Step B3 is specifically included:

B31 when the meaning of a word be not it is multiple, then directly using the meaning of a word as standard language corresponding with the non-default language phrase Phrase;

B32 is then executed when the meaning of a word is multiple and is judged whether multiple meaning of a word are near synonym, if multiple meaning of a word are nearly justice Word selects frequency of use highest meaning of a word output from multiple meaning of a word, using the meaning of a word as with the non-default language phrase pair The standard language phrase answered;If multiple meaning of a word are not near synonym, semantic understanding is based on context carried out, generative semantics understand knot Fruit;Determine that the meaning of a word of the non-default language phrase, the meaning of a word are and the non-default language based on the semantic understanding result The corresponding standard language phrase of words group;

It should be pointed out that in another embodiment, if multiple meaning of a word are not near synonym, based on context carrying out language Reason and good sense solution, generative semantics understand result;According to the volume of the non-default language phrase to judge emotion information;Based on institute's predicate Reason and good sense solution result and emotion information determine the meaning of a word of the non-default language phrase, using the meaning of a word as with the non-default language The corresponding standard language phrase of phrase;

C is located at the location information in the first voice messaging for the standard speech words according to the non-default language phrase of record Group is back to the position in first voice messaging, generates received pronunciation information;

The received pronunciation information is converted text information by D.

3) if so, i.e. described first voice messaging does not include non-default language;It then directly executes and believes first voice Breath is converted into text information.

It should be pointed out that user can also directly input text information, at this point, after receiving the text information of user Directly execute step S103.

Step S103: the text information is handled using verbal model;

Here processing include character segmentation, filtering, classification, part of speech analysis, part-of-speech tagging, extraction label at least one Kind.Wherein, character segmentation refers to participle to obtain multiple phrases, and filtering refers to stop words or meaningless word (such as ", ") mistake Filter.Classification includes being divided into Subject, Predicate and Object according to syntax rule;Part of speech analysis includes notional word, participle differentiation;Verb, noun divide;Word Property standard and according to part of speech analyze result markup information, generate label.

Specifically, carrying out character segmentation, stop words or the filtering of meaningless word to text information using verbal model, dividing Class, part-of-speech tagging, extracts label at part of speech analysis, obtains multiple participle phrases.

Step S104: semantic analysis is carried out to treated text information using the deep learning model of building and generates text The semantic understanding result of word information;

When it is implemented, step S104 includes:

(1) unity and coherence in writing up and down is carried out to treated text information using the deep learning models coupling application scenarios of building Solution and semantic disambiguation, generate the semantic understanding result of text information.

Specifically, the step (1) can be realized one of in the following manner:

First way:

1, it is carried out using multiple phrase combination contexts of the deep learning model of building to treated text information Hereafter understand, semantic disambiguation (polysemant);Obtain the semantic results of multiple phrases;

What above-mentioned semantic disambiguation was carried out mainly for the phrase (including standard language phrase) with multiple meanings, with determination The concrete meaning of the phrase with multiple meanings in text information.

2, the semantic results of multiple phrases are compared with the phrase of knowledge mapping respectively, obtain the similar of each phrase Angle value obtains the semantic results of multiple phrases using the highest phrase of similarity value as the semantic results of each phrase;

Here knowledge mapping is the specialized vocabulary of every field and its term vector map of meaning, is with medical domain Example, knowledge mapping can be the term vector map of medical speciality vocabulary and its meaning, be also possible to disease vocabulary and its meaning, shadow The factor of sound, the relevant diagnosis and treatment map for treating element.

3, the semantic results of multiple phrases are combined.

Specifically, the semantic results of multiple phrases to be returned to the original position of text information, can also utilize the standard Language phrase is combined according to Chinese rule, obtains standard sentence;Using preset standard language model to the standard speech Sentence is analyzed, and is understood the meaning of standard sentence, is generated the semantic understanding result of text information.

By being combined according to Chinese rule, that is, determines the Subject, Predicate and Object of text information, prevent upside-down mounting, while facilitating machine Understanding, is established and unified criterion of identification, raising understand speed and efficiency.

The second way:

A carries out up and down multiple phrase combination contexts of treated text information using the deep learning model of building Unity and coherence in writing solution, semantic disambiguation;Obtain the semantic results of multiple phrases;

The semantic results combination knowledge mapping of multiple phrases is analyzed the internal relation and/or logical relation of multiple phrases by b (causality) generates the semantic understanding result of text information.Here internal relation includes corresponding between multiple phrases System, logical relation includes causality.

Such as the information of input is " whom son of the mother of boy A is ", the multiple phrases obtained by above-mentioned steps: male Child is analyzed by internal relation and logical relation of the knowledge mapping to multiple phrases, is inferred It obtains the meaning of text information and as a result, i.e. answer is " A ".

Step S105: the semantic understanding result of text information is exported.

Specifically, step S105 is mainly realized at least one of in the following manner:

Mode one exports the semantic understanding result of text information in the form of text;

Mode two exports the semantic understanding result of text information in the form of received pronunciation;

Specifically, transferring vocal print using received pronunciation model and being converted to received pronunciation and export;

Further, received pronunciation can also be subjected to two times transfer, is converted to the output of the second voice messaging, above-mentioned the Two voice messagings can be dialect or other language, improve user experience and cordiality degree;

Mode three exports the semantic understanding result of text information in the form of picture.

Mode four exports the semantic understanding result of text information in the form of video.

Mode five exports the semantic understanding result of text information in the form of hyperlink.

Mode six even two or more exports any two kinds in aforesaid way in combination.Such as picture and text are defeated Out or language and characters export etc..

Semantic analysis provided in an embodiment of the present invention, comprising: receive the first voice messaging of user's input;By first Voice messaging is converted into text information;Text information is handled using verbal model;Utilize the deep learning model of building The semantic understanding result that semantic analysis generates text information is carried out to treated text information;By the semanteme reason of text information Solution result is exported.Therefore, technical solution provided in an embodiment of the present invention can be alleviated existing in the prior art traditional Semantic analysis cannot systematically carry out the technical issues of complicated semantic analysis, propose a kind of semantic analysis of systematization Scheme can be improved the accuracy of the semantic analysis of complicated voice input, promote the development of human-computer interaction intelligent.

Embodiment two:

As shown in Fig. 2, on the basis of example 1, present invention implementation provides another semantic analysis, with reality The difference for applying example one is, this method further include:

Step S201: deep learning model is constructed based on neural network;

Wherein, neural network can be convolutional neural networks (Convolutional Neural Network, CNN), depth Spend neural network (Deep Neural Networks, DNN), artificial neural network etc..

Specifically, being trained by mass data to artificial neural network, building obtains deep learning model.

Step S202: it in the training process and/or application process of deep learning model, is assisted by manual intervention Label, to improve the understanding accuracy of participle.

After the step S101 of embodiment one, this method can also include:

Step S203: record backup is carried out to first voice messaging and carries out reduction of speed processing;

Step S204: make pauses in reading unpunctuated ancient writings to the first voice messaging;

Specifically, can based on end-point detection to the first voice messaging into punctuate.

Semantic analysis provided in an embodiment of the present invention, handle and make pauses in reading unpunctuated ancient writings by manual intervention, reduction of speed and etc., have Help improve the accuracy of semantic analysis, while facilitating Record Comparison, deep learning model is corrected, is fed back, improves deep Spend the applicability of learning model.

Embodiment three:

As shown in figure 3, the embodiment of the invention provides a kind of semantic analysis devices, comprising:

Receiving module 301, for receiving the first voice messaging of user's input;

Conversion module 302, for converting text information for first voice messaging;

Processing module 303, for being handled using verbal model the text information;

Analysis module 304 carries out semantic point to treated text information for the deep learning model using building Analysis generates the semantic understanding result of text information;

Output module 305, for exporting the semantic understanding result of text information.

Further, the conversion module 302 is specifically used for: judging whether first voice messaging is standard speech message Breath;If it is not, first voice messaging is then converted to received pronunciation information;Text letter is converted by the received pronunciation information Breath.

Further, the processing module 303 is specifically used for: using verbal model to text information carry out character segmentation, Filtering, part of speech analysis, part-of-speech tagging, extracts label at classification, obtains multiple participle phrases.

Further, the analysis module 304 is specifically used for: utilizing the deep learning models coupling application scenarios pair of building Treated text information carries out context understanding and semantic disambiguation, generates the semantic understanding result of text information.Specifically, Context understanding, language are carried out to multiple phrase combination contexts of treated text information using the deep learning model of building Justice disambiguates;Obtain the semantic results of multiple phrases;The semantic results of multiple phrases are compared with the phrase of knowledge mapping respectively It is right, the similarity value of each phrase is obtained, using the highest phrase of similarity value as the semantic results of each phrase, is obtained multiple The semantic results of phrase;The semantic results of multiple phrases are combined, the semantic understanding result of text information is generated;Alternatively, Context understanding, language are carried out to multiple phrase combination contexts of treated text information using the deep learning model of building Justice disambiguates;Obtain the semantic results of multiple phrases;By the semantic results combination knowledge mapping of multiple phrases, multiple phrases are analyzed Internal relation and/or logical relation generate the semantic understanding result of text information.

Further, the output module 305 is specifically used for: by the semantic understanding result of text information in the form of text It is exported;And/or the semantic understanding result of text information is exported in the form of received pronunciation;And/or by text The semantic understanding result of information is exported in a pattern.

Further, which can also include: training module 306, in deep learning model training process and/ Or in application process, aid mark is carried out by manual intervention, to improve the understanding accuracy of participle.

Semantic analysis device provided in an embodiment of the present invention has identical with semantic analysis provided by the above embodiment Technical characteristic reach identical technical effect so also can solve identical technical problem.

The technical effect and preceding method embodiment phase of device provided by the embodiment of the present invention, realization principle and generation Together, to briefly describe, Installation practice part does not refer to place, can refer to corresponding contents in preceding method embodiment.

It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description It with the specific work process of device, can refer to corresponding processes in the foregoing method embodiment, details are not described herein.

It should be noted that above-mentioned various models, including dialect model, verbal model, deep learning model It can be trained and be generated by neural network learning Deng, specifically be referred to the learning process of neural network, details are not described herein.

Referring to fig. 4, the embodiment of the present invention also provides a kind of electronic equipment 100, comprising: processor 40, memory 41, bus 42 and communication interface 43, the processor 40, communication interface 43 and memory 41 are connected by bus 42;Processor 40 is for holding The executable module stored in line storage 41, such as computer program.

Wherein, memory 41 may include high-speed random access memory (RAM, Random Access Memory), It may further include non-volatile memory (non-volatile memory), for example, at least a magnetic disk storage.By extremely A few communication interface 43 (can be wired or wireless) is realized logical between the system network element and at least one other network element Letter connection, can be used internet, wide area network, local network, Metropolitan Area Network (MAN) etc..

Bus 42 can be isa bus, pci bus or eisa bus etc..The bus can be divided into address bus, data Bus, control bus etc..Only to be indicated with a four-headed arrow convenient for indicating, in Fig. 4, it is not intended that an only bus or A type of bus.

Wherein, memory 41 is for storing program 401, and the processor 40 is after receiving and executing instruction, described in execution Program 401, method performed by the device that the stream process that aforementioned any embodiment of the embodiment of the present invention discloses defines can be applied It is realized in processor 40, or by processor 40.

Processor 40 may be a kind of IC chip, the processing capacity with signal.During realization, above-mentioned side Each step of method can be completed by the integrated logic circuit of the hardware in processor 40 or the instruction of software form.Above-mentioned Processor 40 can be general processor, including central processing unit (Central Processing Unit, abbreviation CPU), network Processor (Network Processor, abbreviation NP) etc.;It can also be digital signal processor (Digital Signal Processing, abbreviation DSP), specific integrated circuit (Application Specific Integrated Circuit, referred to as ASIC), ready-made programmable gate array (Field-Programmable Gate Array, abbreviation FPGA) or other are programmable Logical device, discrete gate or transistor logic, discrete hardware components.It may be implemented or execute in the embodiment of the present invention Disclosed each method, step and logic diagram.General processor can be microprocessor or the processor is also possible to appoint What conventional processor etc..The step of method in conjunction with disclosed in the embodiment of the present invention, can be embodied directly in hardware decoding processing Device executes completion, or in decoding processor hardware and software module combination execute completion.Software module can be located at Machine memory, flash memory, read-only memory, programmable read only memory or electrically erasable programmable memory, register etc. are originally In the storage medium of field maturation.The storage medium is located at memory 41, and processor 40 reads the information in memory 41, in conjunction with Its hardware completes the step of above method.

The embodiment of the invention also provides a kind of computers of non-volatile program code that can be performed with processor can Medium is read, said program code makes the processor execute any method of previous embodiment.

Unless specifically stated otherwise, the opposite step of the component and step that otherwise illustrate in these embodiments, digital table It is not limit the scope of the invention up to formula and numerical value.

It should also be noted that similar label and letter indicate similar terms in following attached drawing, therefore, once a certain Xiang Yi It is defined in a attached drawing, does not then need that it is further defined and explained in subsequent attached drawing.

The flow chart and block diagram in the drawings show the system of multiple embodiments according to the present invention, method and computer journeys The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation A part of one module, section or code of table, a part of the module, section or code include one or more use The executable instruction of the logic function as defined in realizing.It should also be noted that in some implementations as replacements, being marked in box The function of note can also occur in a different order than that indicated in the drawings.For example, two continuous boxes can actually base Originally it is performed in parallel, they can also be executed in the opposite order sometimes, and this depends on the function involved.It is also noted that It is the combination of each box in block diagram and or flow chart and the box in block diagram and or flow chart, can uses and execute rule The dedicated hardware based system of fixed function or movement is realized, or can use the group of specialized hardware and computer instruction It closes to realize.

In addition, in the description of the embodiment of the present invention unless specifically defined or limited otherwise, term " installation ", " phase Even ", " connection " shall be understood in a broad sense, for example, it may be being fixedly connected, may be a detachable connection, or be integrally connected;It can To be mechanical connection, it is also possible to be electrically connected;It can be directly connected, can also can be indirectly connected through an intermediary Connection inside two elements.For the ordinary skill in the art, above-mentioned term can be understood at this with concrete condition Concrete meaning in invention.

In the description of the present invention, it should be noted that term " center ", "upper", "lower", "left", "right", "vertical", The orientation or positional relationship of the instructions such as "horizontal", "inner", "outside" be based on the orientation or positional relationship shown in the drawings, merely to Convenient for description the present invention and simplify description, rather than the device or element of indication or suggestion meaning must have a particular orientation, It is constructed and operated in a specific orientation, therefore is not considered as limiting the invention.In addition, term " first ", " second ", " third " is used for descriptive purposes only and cannot be understood as indicating or suggesting relative importance.

The computer program product of semantic analysis is carried out provided by the embodiment of the present invention, including stores processor The computer readable storage medium of executable non-volatile program code, the instruction that said program code includes can be used for executing Previous methods method as described in the examples, specific implementation can be found in embodiment of the method, and details are not described herein.

It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description, The specific work process of device and unit, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.

In several embodiments provided herein, it should be understood that disclosed systems, devices and methods, it can be with It realizes by another way.The apparatus embodiments described above are merely exemplary, for example, the division of the unit, Only a kind of logical function partition, there may be another division manner in actual implementation, in another example, multiple units or components can To combine or be desirably integrated into another system, or some features can be ignored or not executed.Another point, it is shown or beg for The mutual coupling, direct-coupling or communication connection of opinion can be through some communication interfaces, device or unit it is indirect Coupling or communication connection can be electrical property, mechanical or other forms.

The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme 's.

It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit It is that each unit physically exists alone, can also be integrated in one unit with two or more units.

It, can be with if the function is realized in the form of SFU software functional unit and when sold or used as an independent product It is stored in the executable non-volatile computer-readable storage medium of a processor.Based on this understanding, of the invention Technical solution substantially the part of the part that contributes to existing technology or the technical solution can be with software in other words The form of product embodies, which is stored in a storage medium, including some instructions use so that One computer equipment (can be personal computer, server or the network equipment etc.) executes each embodiment institute of the present invention State all or part of the steps of method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (ROM, Read- Only Memory), random access memory (RAM, Random Access Memory), magnetic or disk etc. are various can be with Store the medium of program code.

Finally, it should be noted that embodiment described above, only a specific embodiment of the invention, to illustrate the present invention Technical solution, rather than its limitations, scope of protection of the present invention is not limited thereto, although with reference to the foregoing embodiments to this hair It is bright to be described in detail, those skilled in the art should understand that: anyone skilled in the art In the technical scope disclosed by the present invention, it can still modify to technical solution documented by previous embodiment or can be light It is readily conceivable that variation or equivalent replacement of some of the technical features;And these modifications, variation or replacement, do not make The essence of corresponding technical solution is detached from the spirit and scope of technical solution of the embodiment of the present invention, should all cover in protection of the invention Within the scope of.Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims (9)

1. a kind of semantic analysis characterized by comprising
Receive the first voice messaging of user's input;
Text information is converted by first voice messaging;It is described to convert text information for first voice messaging, tool Body includes: to judge whether first voice messaging is received pronunciation information;If it is not, then first voice messaging is converted to Received pronunciation information;Text information is converted by the received pronunciation information;Wherein, described if it is not, then by first voice Information is converted to received pronunciation information, comprising:
If it is determined that first voice messaging is not for received pronunciation information, using dialect model to first voice Information is extracted, is divided, and obtains at least one non-default language phrase;And it records non-default language phrase and is located at the first voice Location information in information;
The word of the non-default language phrase is judged according to position of the non-default language phrase in the voice messaging Property;
Phrase, which is searched, based on the part of speech corresponds to table to obtain the meaning of a word corresponding with the non-default language phrase;
When the meaning of a word is multiple, if multiple meaning of a word are not near synonym, semantic understanding is based on context carried out, generative semantics understand As a result;According to the volume of the non-default language phrase to judge emotion information;Believed based on the semantic understanding result and emotion Breath determines the meaning of a word of the non-default language phrase, using the meaning of a word as standard language corresponding with the non-default language phrase Phrase;
It is located at the location information in the first voice messaging according to the non-default language phrase of record to return the standard language phrase It is back to the position in first voice messaging, generates received pronunciation information;
The text information is handled using verbal model;
The semanteme that semantic analysis generates text information is carried out to treated text information using the deep learning model of building Understand result;
The semantic understanding result of text information is exported.
2. the method according to claim 1, wherein it is described using verbal model to the text information at Reason, specifically includes:
Character segmentation is carried out to text information using verbal model, filtering, classification, part of speech analysis, part-of-speech tagging, extracts label, Obtain multiple participle phrases.
3. the method according to claim 1, wherein the deep learning model using building is to by handling Text information carry out semantic analysis generate text information semantic understanding as a result, specifically including:
Context understanding and language are carried out to treated text information using the deep learning models coupling application scenarios of building Justice disambiguates, and generates the semantic understanding result of text information.
4. according to the method described in claim 3, it is characterized in that, the deep learning models coupling applied field using building Scape carries out context understanding and semantic disambiguation to treated text information, generates the semantic understanding of text information as a result, tool Body includes:
Unity and coherence in writing up and down is carried out to multiple phrase combination contexts of treated text information using the deep learning model of building Solution, semantic disambiguation;Obtain the semantic results of multiple phrases;
The semantic results of multiple phrases are compared with the phrase of knowledge mapping respectively, obtain the similarity value of each phrase, Using the highest phrase of similarity value as the semantic results of each phrase, the semantic results of multiple phrases are obtained;
The semantic results of multiple phrases are combined, the semantic understanding result of text information is generated.
5. according to the method described in claim 3, it is characterized in that, the deep learning models coupling applied field using building Scape carries out context understanding and semantic disambiguation to treated text information, generates the semantic understanding of text information as a result, tool Body includes:
Unity and coherence in writing up and down is carried out to multiple phrase combination contexts of treated text information using the deep learning model of building Solution, semantic disambiguation;Obtain the semantic results of multiple phrases;
By the semantic results combination knowledge mapping of multiple phrases, the internal relation and/or logical relation of multiple phrases are analyzed, is generated The semantic understanding result of text information.
6. the method according to claim 1, wherein the semantic understanding result progress by text information is defeated Out, it specifically includes:
The semantic understanding result of text information is exported in the form of text;
And/or the semantic understanding result of text information is exported in the form of received pronunciation;
And/or the semantic understanding result of text information is exported in the form of picture;
And/or the semantic understanding result of text information is exported in the form of video;
And/or the semantic understanding result of text information is exported in the form of hyperlink.
7. the method according to claim 1, wherein further include:
In the training process and/or application process of the deep learning model, aid mark is carried out by manual intervention, to mention The understanding accuracy of height participle.
8. a kind of semantic analysis device characterized by comprising
Receiving module, for receiving the first voice messaging of user's input;
Conversion module, for converting text information for first voice messaging;Conversion module is for judging first language Whether message breath is received pronunciation information;If it is not, first voice messaging is then converted to received pronunciation information;By the mark Quasi- voice messaging is converted into text information;Wherein, described if it is not, first voice messaging is then converted to standard speech message Breath, comprising:
If it is determined that first voice messaging is not for received pronunciation information, using dialect model to first voice Information is extracted, is divided, and obtains at least one non-default language phrase;And it records non-default language phrase and is located at the first voice Location information in information;
The word of the non-default language phrase is judged according to position of the non-default language phrase in the voice messaging Property;
Phrase, which is searched, based on the part of speech corresponds to table to obtain the meaning of a word corresponding with the non-default language phrase;
When the meaning of a word is multiple, if multiple meaning of a word are not near synonym, semantic understanding is based on context carried out, generative semantics understand As a result;According to the volume of the non-default language phrase to judge emotion information;Believed based on the semantic understanding result and emotion Breath determines the meaning of a word of the non-default language phrase, using the meaning of a word as standard language corresponding with the non-default language phrase Phrase;
It is located at the location information in the first voice messaging according to the non-default language phrase of record to return the standard language phrase It is back to the position in first voice messaging, generates received pronunciation information;
Processing module, for being handled using verbal model the text information;
Analysis module carries out semantic analysis to treated text information for the deep learning model using building and generates text The semantic understanding result of word information;
Output module, for exporting the semantic understanding result of text information.
9. a kind of electronic equipment, including memory, processor and it is stored on the memory and can transports on the processor Capable computer program, which is characterized in that the processor realizes the claims 1 to 7 when executing the computer program The step of described in any item methods.
CN201810534587.2A 2018-05-29 2018-05-29 Semantic analysis, device and electronic equipment CN108806671B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810534587.2A CN108806671B (en) 2018-05-29 2018-05-29 Semantic analysis, device and electronic equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810534587.2A CN108806671B (en) 2018-05-29 2018-05-29 Semantic analysis, device and electronic equipment

Publications (2)

Publication Number Publication Date
CN108806671A CN108806671A (en) 2018-11-13
CN108806671B true CN108806671B (en) 2019-06-28

Family

ID=64089230

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810534587.2A CN108806671B (en) 2018-05-29 2018-05-29 Semantic analysis, device and electronic equipment

Country Status (1)

Country Link
CN (1) CN108806671B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2020211006A1 (en) * 2019-04-17 2020-10-22 深圳市欢太科技有限公司 Speech recognition method and apparatus, storage medium and electronic device
CN110111795A (en) * 2019-04-23 2019-08-09 维沃移动通信有限公司 A kind of method of speech processing and terminal device
CN110246496A (en) * 2019-07-01 2019-09-17 珠海格力电器股份有限公司 Audio recognition method, system, computer equipment and storage medium

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103577148A (en) * 2013-11-28 2014-02-12 南京奇幻通信科技有限公司 Voice reading method and device
CN102968409B (en) * 2012-11-23 2015-09-09 海信集团有限公司 Intelligent human-machine interaction semantic analysis and interactive system
CN106023985A (en) * 2016-05-19 2016-10-12 北京捷通华声科技股份有限公司 Linguistic model training method and system and speech recognition system
CN106202301A (en) * 2016-07-01 2016-12-07 武汉泰迪智慧科技有限公司 A kind of intelligent response system based on degree of depth study
CN106448670A (en) * 2016-10-21 2017-02-22 竹间智能科技(上海)有限公司 Dialogue automatic reply system based on deep learning and reinforcement learning
CN107464566A (en) * 2017-09-21 2017-12-12 百度在线网络技术(北京)有限公司 Audio recognition method and device
CN107679039A (en) * 2017-10-17 2018-02-09 北京百度网讯科技有限公司 The method and apparatus being intended to for determining sentence
CN107729309A (en) * 2016-08-11 2018-02-23 中兴通讯股份有限公司 A kind of method and device of the Chinese semantic analysis based on deep learning
CN108022586A (en) * 2017-11-30 2018-05-11 百度在线网络技术(北京)有限公司 Method and apparatus for controlling the page

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102968409B (en) * 2012-11-23 2015-09-09 海信集团有限公司 Intelligent human-machine interaction semantic analysis and interactive system
CN103577148A (en) * 2013-11-28 2014-02-12 南京奇幻通信科技有限公司 Voice reading method and device
CN106023985A (en) * 2016-05-19 2016-10-12 北京捷通华声科技股份有限公司 Linguistic model training method and system and speech recognition system
CN106202301A (en) * 2016-07-01 2016-12-07 武汉泰迪智慧科技有限公司 A kind of intelligent response system based on degree of depth study
CN107729309A (en) * 2016-08-11 2018-02-23 中兴通讯股份有限公司 A kind of method and device of the Chinese semantic analysis based on deep learning
CN106448670A (en) * 2016-10-21 2017-02-22 竹间智能科技(上海)有限公司 Dialogue automatic reply system based on deep learning and reinforcement learning
CN107464566A (en) * 2017-09-21 2017-12-12 百度在线网络技术(北京)有限公司 Audio recognition method and device
CN107679039A (en) * 2017-10-17 2018-02-09 北京百度网讯科技有限公司 The method and apparatus being intended to for determining sentence
CN108022586A (en) * 2017-11-30 2018-05-11 百度在线网络技术(北京)有限公司 Method and apparatus for controlling the page

Also Published As

Publication number Publication date
CN108806671A (en) 2018-11-13

Similar Documents

Publication Publication Date Title
Williams et al. A broad-coverage challenge corpus for sentence understanding through inference
US10262062B2 (en) Natural language system question classifier, semantic representations, and logical form templates
Awal et al. A global learning approach for an online handwritten mathematical expression recognition system
CN104050256B (en) Initiative study-based questioning and answering method and questioning and answering system adopting initiative study-based questioning and answering method
Carreras et al. Introduction to the CoNLL-2005 shared task: Semantic role labeling
Yang et al. Extracting opinion expressions with semi-markov conditional random fields
CN104361127B (en) The multilingual quick constructive method of question and answer interface based on domain body and template logic
US9805718B2 (en) Clarifying natural language input using targeted questions
CN104268603B (en) Intelligent marking method and system for text objective questions
US20160328467A1 (en) Natural language question answering method and apparatus
CN105096935B (en) A kind of pronunciation inputting method, device and system
JP2015094848A (en) Information processor, information processing method and program
RU2607975C2 (en) Constructing corpus of comparable documents based on universal measure of similarity
US20170286397A1 (en) Predictive Embeddings
Kübler et al. Dependency parsing
CN103336766B (en) Short text garbage identification and modeling method and device
CN107016994B (en) Voice recognition method and device
JP4427500B2 (en) Semantic analysis device, semantic analysis method, and semantic analysis program
CN107992543B (en) Question-answer interaction method and device, computer equipment and computer readable storage medium
CN107220235B (en) Speech recognition error correction method and device based on artificial intelligence and storage medium
Shen et al. Straight to the tree: Constituency parsing with neural syntactic distance
CN106777013B (en) Conversation management method and device
CN106534548B (en) Voice error correction method and device
CN106227721B (en) Chinese Prosodic Hierarchy forecasting system
US10068174B2 (en) Hybrid approach for developing, optimizing, and executing conversational interaction applications

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