CN110956962A - Reply information determination method, device and equipment for vehicle-mounted robot - Google Patents
Reply information determination method, device and equipment for vehicle-mounted robot Download PDFInfo
- Publication number
- CN110956962A CN110956962A CN201910990232.9A CN201910990232A CN110956962A CN 110956962 A CN110956962 A CN 110956962A CN 201910990232 A CN201910990232 A CN 201910990232A CN 110956962 A CN110956962 A CN 110956962A
- Authority
- CN
- China
- Prior art keywords
- sample
- participle
- voice
- file
- matrix
- 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
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; 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 OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/26—Speech to text systems
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/28—Constructional details of speech recognition systems
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L2015/223—Execution procedure of a spoken command
Abstract
The invention discloses a method, a device and equipment for determining reply information of a vehicle-mounted robot. The method comprises the following steps: performing word segmentation on the received user voice to obtain a voice word segmentation set of the user voice; forming a voice input file of the user voice according to the voice word set and a pre-formed phrase list dictionary; determining target semantic parameters corresponding to each row in the voice input file, and searching the target semantic parameters in a predetermined voice reply template, wherein the voice reply template is determined based on a given sample corpus and comprises at least one standard semantic parameter and a standard reply participle sequence arranged according to a corresponding statement expression rule; and determining reply information corresponding to the user voice according to the search result. According to the technical scheme of the embodiment of the invention, the effect of accurately matching the user voice with the reply information in the voice reply template and improving the accuracy of the determination of the reply information of the vehicle-mounted robot is achieved.
Description
Technical Field
The embodiment of the invention relates to the technical field of intelligent robots, in particular to a method, a device and equipment for determining reply information of a vehicle-mounted robot.
Background
With the rapid development of intelligent networking in the vehicle manufacturing industry and the internet industry, the artificial intelligence technology has become one of the indispensable subjects in the field of intelligent networking, and research and development of intelligent chat robots in a vehicle-mounted scene has also become a development hotspot in which the industry is imperative.
At present, the word vector construction of the intelligent chat robot based on the seq2seq model mainly adopts a word bag model vector representation method. The vector representation method is a simplified expression model under natural language processing and information retrieval, and is characterized in that a text, a paragraph or a document is regarded as a disordered vocabulary set, grammar and word sequence are omitted, each word in the text is counted, and word vector index content is represented by the occurrence frequency of the words in a list.
However, the vector representation method does not keep the order of words in the original sentence, only takes each word group as individual statistics, and in the practical question answering application, the word vectors constructed by the method cannot be directly converted into the sequence recognizable by the computer, and the reply information of the reducible sentence structure cannot be formed, so that the accuracy of determining the reply information is influenced.
Disclosure of Invention
The invention provides a method, a device and equipment for determining reply information of a vehicle-mounted robot, which improve the accuracy of determining the reply information of the vehicle-mounted robot.
In a first aspect, an embodiment of the present invention provides a reply information determination method for a vehicle-mounted robot, including:
performing word segmentation on the received user voice to obtain a voice word segmentation set of the user voice;
forming a voice input file of the user voice according to the voice word set and a pre-formed phrase list dictionary;
determining target semantic parameters corresponding to each row in the voice input file, and searching the target semantic parameters in a predetermined voice reply template, wherein the voice reply template is determined based on a given sample corpus and comprises at least one standard semantic parameter and a standard reply participle sequence arranged according to a corresponding statement expression rule;
and determining reply information corresponding to the user voice according to the search result.
In a second aspect, an embodiment of the present invention further provides a reply information determination apparatus for a vehicle-mounted robot, including:
the voice acquisition module is used for performing word segmentation processing on the received user voice to obtain a voice word segmentation set of the user voice;
the file generation module is used for forming a voice input file of the user voice according to the voice word set and a pre-formed phrase list dictionary;
a template determination module for determining a voice response template based on a given sample corpus;
the parameter searching module is used for determining target semantic parameters corresponding to each row in the voice input file and searching the target semantic parameters in a predetermined voice reply template;
and the information determining module is used for determining the reply information corresponding to the user voice according to the search result.
In a third aspect, an embodiment of the present invention further provides an apparatus, where the apparatus includes:
one or more processors;
storage means for storing one or more programs;
when the one or more programs are executed by the one or more processors, the one or more processors implement the reply information determination method of the in-vehicle robot as provided in any embodiment of the present invention.
In a fourth aspect, embodiments of the present invention provide a storage medium containing computer-executable instructions that, when executed by a computer processor, are used to perform a reply information determination method for a vehicle-mounted robot as provided in any of the embodiments of the present invention.
The embodiment of the invention obtains the voice word segmentation set of the user voice by word segmentation processing of the received user voice; forming a voice input file of the user voice according to the voice word set and a pre-formed phrase list dictionary; determining target semantic parameters corresponding to each row in the voice input file, and searching the target semantic parameters in a predetermined voice reply template, wherein the voice reply template is determined based on a given sample corpus and comprises at least one standard semantic parameter and a standard reply participle sequence arranged according to a corresponding statement expression rule; and determining reply information corresponding to the user voice according to the search result. And forming a voice input file of the user voice according to the voice word segmentation set and a pre-formed phrase list dictionary, so that the word sequence in the original voice can be reserved, and the inaccuracy of sentence structure reduction is avoided. The target semantic parameters in the voice input file are matched with the standard semantic parameters in the voice reply template, the reply information corresponding to the voice of the user is determined according to the matching result, the problem that the question and the answer are easy to be confused when the voice of the user is matched with the sentence of the reply template is solved, and the accuracy of determining the reply information of the vehicle-mounted robot is improved.
Drawings
Fig. 1 is a flowchart of a reply information determination method of a vehicle-mounted robot according to a first embodiment of the present invention;
fig. 2 is a flowchart of a reply information determination method of a vehicle-mounted robot according to a second embodiment of the present invention;
FIG. 3 is a flowchart of determining a voice response template in a second embodiment of the present invention;
fig. 4 is a schematic structural diagram of a reply information determination device of a vehicle-mounted robot in a third embodiment of the present invention;
fig. 5 is a schematic structural diagram of an apparatus according to a fourth embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
Example one
Fig. 1 is a flowchart of a reply information determination method for a vehicle-mounted robot according to an embodiment of the present invention, where the present embodiment is applicable to a case where the vehicle-mounted robot determines reply information, the method may be executed by a reply information determination device of the vehicle-mounted robot, the reply information determination device of the vehicle-mounted robot may be implemented by software and/or hardware, and the reply information determination device of the vehicle-mounted robot may be configured on a computing device, and specifically includes the following steps:
and 110, performing word segmentation on the received user voice to obtain a voice word segmentation set of the user voice.
The user voice can be understood as voice information collected by a sound collection device, and the voice information can contain one or more sentences, and can be specifically represented by a group of character strings.
The word segmentation process can be understood as Chinese word segmentation, that is, a process of recombining continuous character sequences into word sequences according to a certain specification, and specifically, the word segmentation process can adopt a word segmentation method based on character string matching, a word segmentation method based on understanding, a word segmentation method based on statistics, and the like.
Specifically, the received user voice signal can be subjected to pre-emphasis processing, the high-frequency part of the voice signal is emphasized, the high-frequency resolution is increased, meanwhile, the received user voice signal can be subjected to windowing processing, regular background noise is eliminated, the background noise is converted into a character string mode, the converted user voice is subjected to word segmentation processing, and after the processing, pause word removal and low-frequency word removal processing can be carried out to form a voice word segmentation set of the user voice.
The received user voice is processed by word segmentation, the user voice can be converted into standard voice sequences, the computer can conveniently count and recognize the word sequences, the problem that the received user voice is difficult to recognize is avoided, and the reading efficiency of the subsequent computer on the received user voice is improved.
And step 120, forming a voice input file of the user voice according to the voice word set and a pre-formed phrase list dictionary.
The phrase list dictionary may be understood as that the phrases in the sample corpus are sequentially arranged and numbered according to a preset ordering method according to a preset sample corpus to generate a phrase set with numbers corresponding to the phrases one by one, specifically, the preset ordering method may adopt ordering according to the frequency of occurrence of the phrases in the sample corpus, ordering according to the weight of the phrases in a sample lexicon, ordering according to the sequence of occurrence of the phrases in the sample corpus, and the like.
Specifically, the phrases in the voice segmentation set are assigned according to the numbers corresponding to the phrases in the phrase list dictionary, that is, the voice segmentation set can be represented as a matrix consisting of the numbers, and the matrix is used as the voice input file of the voice of the user.
And forming a voice input file of the user voice according to the voice word segmentation set and a pre-formed phrase list dictionary, so that the word sequence in the original voice can be reserved, and the inaccuracy of sentence structure reduction is avoided.
The semantic parameters can be understood as mathematical expressions which comprehensively embody grammatical information, semantic information and pragmatic information in a sentence, and can be specifically represented by a group of ordered numbers.
The voice response template may be understood as a matrix file containing a plurality of semantic parameters and response sentences corresponding to the semantic parameters one by one, wherein the response sentences may be represented by a standard response complex word sequence.
Specifically, firstly, sample question sentences and sample answer sentences in a sample corpus set are extracted, a sample question sentence file, a sample answer sentence file and a sample collation file are constructed, semantic parameters and answer sentences in the sample answer sentence file are matched according to the semantic parameters corresponding to the question sentences in the sample question sentence file, the matching conditions of the semantic parameters and the answer sentences are collated according to the sample collation file, and a voice answer template containing standard semantic parameters and corresponding standard answer participle sequences is generated. And extracting target semantic parameters corresponding to each sentence in the voice input file, and performing matching search on the extracted target semantic parameters and the standard semantic parameters in the voice reply template to obtain a search result.
And step 140, determining reply information corresponding to the user voice according to the search result.
The reply information of the user voice is a standard reply word segmentation sequence or a preset reply sentence which is obtained by matching the semantic parameters in the voice reply template.
Specifically, if the search result shows that the target semantic parameter is matched with the standard semantic parameter in the voice reply template, the question provided by the user is in the range of the voice reply template, and at the moment, the standard reply participle sequence corresponding to the standard semantic parameter is output as the reply information of the voice of the user; otherwise, the question posed by the user is indicated to be beyond the range of the voice response template, and at the moment, a preset response sentence is output as the response information of the voice of the user.
According to the technical scheme of the embodiment, the received user voice is subjected to word segmentation processing to obtain a voice word segmentation set of the user voice; forming a voice input file of the user voice according to the voice word set and a pre-formed phrase list dictionary; determining target semantic parameters corresponding to each row in the voice input file, and searching the target semantic parameters in a predetermined voice reply template, wherein the voice reply template is determined based on a given sample corpus and comprises at least one standard semantic parameter and a standard reply participle sequence arranged according to a corresponding statement expression rule; and determining reply information corresponding to the user voice according to the search result. And forming a voice input file of the user voice according to the voice word segmentation set and a pre-formed phrase list dictionary, so that the word sequence in the original voice can be reserved, and the inaccuracy of sentence structure reduction is avoided. The target semantic parameters in the voice input file are matched with the standard semantic parameters in the voice reply template, the reply information corresponding to the voice of the user is determined according to the matching result, the problem that the question and the answer are easy to be confused when the voice of the user is matched with the sentence of the reply template is solved, and the accuracy of determining the reply information of the vehicle-mounted robot is improved.
Example two
Fig. 2 is a flowchart of a reply information determination method for a vehicle-mounted robot according to a second embodiment of the present invention. The technical scheme of the embodiment is further refined on the basis of the technical scheme, and specifically comprises the following steps:
The sample corpus can be a sentence set conforming to an application environment determined according to an actual application situation, such as a control corpus, a music corpus, a radio corpus, a telephone corpus, a navigation corpus, a video corpus, a charging corpus, a weather corpus, a stock corpus, a hotel corpus, and the like. The sample corpus can be in a question-answer form, i.e. a question sentence and an answer sentence can be distinguished according to the number of lines.
And 203, counting the occurrence frequency of each participle in the corpus participle set.
Specifically, each participle in the corpus participle set can be regarded as a phrase form, and the occurrence frequency of each phrase in the whole corpus participle set can be calculated through a traversal method and is used as the occurrence frequency of each participle.
And 204, arranging corresponding corpus participles according to the occurrence frequency from large to small, extracting the corpus participles in the preset number and corresponding arrangement serial numbers to form a phrase list dictionary.
Specifically, the words in the corpus word set are arranged in a descending order according to the occurrence frequency of the words, if the occurrence frequency of the two words is the same, the words are arranged in a first-to-last order, and after the arrangement is completed, a preset number of corpus words and the corresponding arrangement sequence number are selected to generate a word group list dictionary.
Optionally, the set number may be any positive integer value not less than 95% of the number of the corpus participles in the corpus participle set.
And step 205, initializing and constructing a voice input matrix.
The voice word segmentation set comprises at least one line of word segmentation processing first word segmentation sequences formed after the user voice is processed.
The first word segmentation sequence can be understood as user voice after word segmentation processing of one sentence in the voice word segmentation set, and the user voice after word segmentation processing is sequentially stored in each line of the voice word segmentation set by taking the sentence as a unit.
The number of rows of the voice input matrix is equal to the number of rows of a first word segmentation sequence contained in the voice word segmentation set, and the number of columns of the voice input matrix is equal to the number of word segmentation of a first word segmentation contained in the longest first word segmentation sequence in the voice word segmentation set.
The preset code value can adopt zero value, pad character or other fixed character set according to user habit.
Specifically, each first word in the voice word set is assigned, a phrase list dictionary is traversed, and if the first word exists in the phrase list dictionary, the corresponding arrangement serial number of the first word in the phrase list dictionary is used as a target coding value; if the first segmentation word does not exist in the phrase list dictionary, the preset coding value is used as the target coding value of the first segmentation word.
And step 208, taking the assigned voice input matrix as a voice input file of the user voice.
The target semantic parameters can be understood as mathematical expressions comprehensively embodied by grammar information, semantic information and pragmatic information in sentences corresponding to each line in the voice input file.
Further, FIG. 3 provides a flow chart for determining a voice response template. As shown in fig. 3, the sample corpus includes at least one group of sample question-and-answer sentences, each group of sample question-and-answer sentences includes a row of sample question-and-answer sentences and a row of associated sample answer-and-answer sentences; accordingly, the process of determining the voice response template includes the steps of:
Specifically, forming a sample question file according to the sample question set includes:
and performing word segmentation on the sample question set to obtain a question word set comprising at least one line of second word segmentation sequence.
And initializing to construct a sample question matrix, wherein the row number of the sample question matrix is equal to the row number of a second participle sequence contained in the question participle set, and the column number of the sample question matrix is equal to the participle number of a second participle contained in the longest second participle sequence in the question participle set.
And aiming at each second participle in the sample question sentence set, taking the corresponding arrangement serial number of the second participle in the phrase list dictionary as a target coding value of the second participle.
And correspondingly writing target coding values of second participles contained in each participle sequence into the sample question matrix by taking the participle sequence as a row unit, and filling empty elements in the sample question matrix with preset coding values.
And taking the sample question matrix after assignment as the sample question file.
Specifically, forming a sample sentence answering file and a sample proofreading file according to the sample sentence answering set and combination sentence expression rule includes:
and performing word segmentation on the sample answer sentence set to obtain an answer sentence word set containing at least one line of third word sequences.
And aiming at each third participle in the sample answer sentence set, taking the corresponding arrangement serial number of the third participle in the phrase list dictionary as a target coding value of the third participle.
And initializing to construct a sample answer sentence matrix, wherein the row number of the sample answer sentence matrix is equal to the row number of a third participle sequence contained in the answer sentence participle set, and the column number of the sample answer sentence matrix is equal to the participle number of a third participle contained in the longest third participle sequence in the answer sentence participle set.
And correspondingly writing target coding values of third participles contained in each participle sequence into the sample answer sentence matrix by taking the participle sequence as a row unit, and filling null elements in the sample answer sentence matrix by adopting preset coding values.
And initializing to construct a sample proofreading matrix, wherein the row number of the sample proofreading matrix is equal to the row number of a third participle sequence contained in the answer sentence participle set, and the column number of the sample proofreading matrix is equal to the participle number of a third participle contained in the longest third participle sequence in the answer sentence participle set.
And correspondingly writing target coding values of third participles contained in each participle sequence into the sample proofreading matrix by taking the participle sequence as a row unit, and filling empty elements in the sample proofreading matrix with preset coding values.
And taking the sample sentence answering matrix after assignment as the sample sentence answering file, and taking the sample proofreading matrix after assignment as the sample proofreading file.
The written sample sentence-answering matrix takes a stop symbol as an end identifier, the written sample proofreading matrix takes a start symbol as an initial identifier, and the stop symbol as an end identifier. Alternatively, the initiator may be go and the rest may be eos.
And 2094, combining the standard semantic parameters with the sample answer sentence file and the sample proofreading file to form a voice answer template containing the standard semantic parameters and corresponding standard answer segmentation sequences.
Specifically, the standard semantic parameters are matched with the sample answer file, and an initial voice answer template corresponding to each participle sequence in the answer file is generated according to each standard semantic parameter.
Matching each participle sequence in the initial voice reply template with the sample proofreading file one by one, replacing participle sequences with different matching results with corresponding participle sequences in the sample proofreading file, taking the replaced participle sequences as the standard reply participle sequences, and generating the voice reply template with the standard semantic parameters corresponding to the standard reply participle sequences.
And step 210, determining reply information corresponding to the user voice according to the search result.
Specifically, if the target semantic parameter is the standard semantic parameter in the voice reply template, the standard reply participle sequence corresponding to the target semantic parameter in the voice reply template is used as the reply information of the user voice; otherwise, the preset reply sentence is used as the reply information of the user voice.
According to the technical scheme of the embodiment, the answer information corresponding to the voice of the user is determined according to the matching result by utilizing the matching of the semantic parameters and the target semantic parameters in the voice answer template containing the semantic parameters and the answer participle sequence, so that the confusion probability of similar semantic question-answer sentences is reduced, and the accuracy of answer information determination is improved.
EXAMPLE III
Fig. 4 is a schematic structural diagram of a reply information determination apparatus for a vehicle-mounted robot according to a third embodiment of the present invention, where the reply information determination apparatus for a vehicle-mounted robot includes: a voice acquisition module 310, a file generation module 320, a template determination module 330, a parameter lookup module 340, and an information determination module 350.
The voice acquiring module 310 is configured to perform word segmentation on the received user voice to obtain a voice word segmentation set of the user voice; a file generating module 320, configured to form a voice input file of the user voice according to the voice word set and a pre-formed phrase list dictionary; a template determination module 330 for determining a voice response template based on a given sample corpus; a parameter searching module 340, configured to determine target semantic parameters corresponding to each line in the voice input file, and search the target semantic parameters in a predetermined voice reply template; and an information determining module 350, configured to determine reply information corresponding to the user voice according to the search result.
According to the technical scheme, the problem that the similar question sentences and answer sentences are easy to confuse when the user voice is matched with the answer template sentences is solved, and the accuracy of determining the answer information of the vehicle-mounted robot is improved.
Optionally, the file generating module 320 includes:
the phrase list dictionary forming unit is used for carrying out word segmentation on a predetermined sample corpus to obtain a corpus word segmentation set; counting the occurrence frequency of each participle in the corpus participle set; and arranging corresponding corpus participles according to the occurrence frequency from large to small, and extracting the corpus participles in the preset number and corresponding arrangement serial numbers to form a phrase list dictionary.
The input file forming unit is used for initializing and constructing a voice input matrix, wherein the row number of the voice input matrix is equal to the row number of a first participle sequence contained in the voice participle set, and the column number of the voice input matrix is equal to the participle number of a first participle contained in the longest first participle sequence in the voice participle set; aiming at each first participle in the voice participle set, if the first participle exists in the phrase list dictionary, taking a sequence number corresponding to the first participle as a target coding value; otherwise, using a preset coding value as a target coding value of the first word segmentation; correspondingly writing target coding values of first participles contained in each participle sequence into the voice input matrix by taking the participle sequence as a row unit, and filling empty elements in the voice input matrix with the preset coding values; and taking the assigned voice input matrix as a voice input file of the voice of the user.
The voice word segmentation set comprises at least one line of word segmentation processing first word segmentation sequences formed after the user voice is processed.
Optionally, the template determining module 330 includes:
and the sentence set forming unit is used for extracting the sample question sentences and the sample answer sentences in the sample corpus set to form a sample question sentence set and a sample answer sentence set.
And the file forming unit is used for forming a sample question file according to the sample question set and forming a sample answer file and a sample proofreading file according to the sample answer set integrated statement expression rule.
And the semantic parameter determining unit is used for determining standard semantic parameters corresponding to each line of sample participles of the sample question sentence file.
And the reply template forming unit is used for combining each standard semantic parameter according to the sample reply sentence file and the sample proofreading file to form a voice reply template containing each standard semantic parameter and a corresponding standard reply participle sequence.
Optionally, the file forming unit is further configured to perform word segmentation on the sample question set to obtain a question word set including at least one line of the second word segmentation sequence; initializing and constructing a sample question matrix, wherein the row number of the sample question matrix is equal to the row number of a second participle sequence contained in the question participle set, and the column number of the sample question matrix is equal to the participle number of a second participle contained in the longest second participle sequence in the question participle set; aiming at each second participle in the sample question sentence set, taking the corresponding arrangement serial number of the second participle in the phrase list dictionary as a target coding value of the second participle; correspondingly writing target coding values of second participles contained in each participle sequence into the sample question matrix by taking the participle sequence as a row unit, and filling empty elements in the sample question matrix with preset coding values; and taking the sample question matrix after assignment as the sample question file.
Optionally, the file forming unit is further configured to perform word segmentation on the sample answer set to obtain an answer word segmentation set including at least one row of the third entry sequence; aiming at each third participle in the sample answer sentence set, taking the corresponding arrangement serial number of the third participle in the phrase list dictionary as a target coding value of the third participle; initializing and constructing a sample answer sentence matrix, wherein the row number of the sample answer sentence matrix is equal to the row number of a third participle sequence contained in the answer sentence participle set, and the column number of the sample answer sentence matrix is equal to the participle number of a third participle contained in the longest third participle sequence in the answer sentence participle set; correspondingly writing target coding values of third participles contained in each participle sequence into the sample answer sentence matrix by taking the participle sequence as a row unit, and filling null elements in the sample answer sentence matrix with preset coding values; initializing and constructing a sample proofreading matrix, wherein the row number of the sample proofreading matrix is equal to the row number of a third participle sequence contained in the answer sentence participle set, and the column number of the sample proofreading matrix is equal to the participle number of a third participle contained in the longest third participle sequence in the answer sentence participle set; correspondingly writing target coding values of third participles contained in each participle sequence into the sample proofreading matrix by taking the participle sequence as a row unit, and filling empty elements in the sample proofreading matrix with preset coding values; and taking the sample sentence answering matrix after assignment as the sample sentence answering file, and taking the sample proofreading matrix after assignment as the sample proofreading file.
Optionally, the reply template forming unit is further configured to match the standard semantic parameters with the sample reply sentence file, and generate an initial voice reply template in which each standard semantic parameter corresponds to each participle sequence in the reply file; matching each participle sequence in the initial voice reply template with the sample proofreading file one by one, replacing participle sequences with different matching results with corresponding participle sequences in the sample proofreading file, taking the replaced participle sequences as the standard reply participle sequences, and generating the voice reply template with the standard semantic parameters corresponding to the standard reply participle sequences.
Optionally, the information determining module 350 is specifically configured to: if the target semantic parameter is the standard semantic parameter in the voice reply template, taking the standard reply participle sequence corresponding to the target semantic parameter in the voice reply template as the reply information of the user voice; otherwise, the preset reply sentence is used as the reply information of the user voice.
The reply information determining device of the vehicle-mounted robot provided by the embodiment of the invention can execute the reply information determining method of the vehicle-mounted robot provided by any embodiment of the invention, and has corresponding functional modules and beneficial effects of the executing method.
Example four
Fig. 5 is a schematic structural diagram of an apparatus according to a fourth embodiment of the present invention, as shown in fig. 5, the apparatus includes a processor 410, a memory 420, an input device 430, and an output device 440; the number of processors 410 in the device may be one or more, and one processor 4100 is exemplified in fig. 5; the processor 410, the memory 420, the input device 430 and the output device 440 in the apparatus may be connected by a bus or other means, for example, in fig. 5.
The memory 420 serves as a computer-readable storage medium, and may be used for storing software programs, computer-executable programs, and modules, such as program instructions/modules corresponding to the missed fee detection method in the embodiment of the present invention (for example, the voice obtaining module 310, the file generating module 320, the template determining module 330, the parameter searching module 340, and the information determining module 350). The processor 410 executes various functional applications of the device and data processing by running software programs, instructions, and modules stored in the memory 420, that is, implements the reply information determination method of the in-vehicle robot described above.
The memory 420 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required for at least one function; the storage data area may store data created according to the use of the terminal, and the like. Further, the memory 420 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. In some examples, memory 420 may further include memory located remotely from processor 410, which may be connected to devices through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The input device 430 may be used to receive input numeric or character information and generate key signal inputs related to user settings and function control of the computing device, and may include a keyboard and a mouse, etc. The output device 440 may include a display device such as a display screen.
EXAMPLE five
The fifth embodiment of the present invention further provides a storage medium containing computer-executable instructions, which are used for executing a reply information determination method of a vehicle-mounted robot when executed by a computer processor.
Of course, the storage medium containing the computer-executable instructions provided by the embodiments of the present invention is not limited to the method operations described above, and may also perform related operations in a reply information determination method for a vehicle-mounted robot provided by any embodiments of the present invention.
From the above description of the embodiments, it is obvious for those skilled in the art that the present invention can be implemented by software and necessary general hardware, and certainly, can also be implemented by hardware, but the former is a better embodiment in many cases. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which may be stored in a computer-readable storage medium, such as a floppy disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a FLASH Memory (FLASH), a hard disk or an optical disk of a computer, and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device) to execute the methods according to the embodiments of the present invention.
It should be noted that, in the embodiment of the reply information determination apparatus for a vehicle-mounted robot, the units and modules included in the apparatus are merely divided according to the functional logic, but are not limited to the above division as long as the corresponding functions can be realized; in addition, specific names of the functional units are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present invention.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (10)
1. A reply information determination method for a vehicle-mounted robot, comprising:
performing word segmentation on the received user voice to obtain a voice word segmentation set of the user voice;
forming a voice input file of the user voice according to the voice word set and a pre-formed phrase list dictionary;
determining target semantic parameters corresponding to each row in the voice input file, and searching the target semantic parameters in a predetermined voice reply template, wherein the voice reply template is determined based on a given sample corpus and comprises at least one standard semantic parameter and a standard reply participle sequence arranged according to a corresponding statement expression rule;
and determining reply information corresponding to the user voice according to the search result.
2. The method of claim 1, wherein forming the phrase list dictionary comprises:
performing word segmentation on a predetermined sample corpus to obtain a corpus word segmentation set;
counting the occurrence frequency of each participle in the corpus participle set;
and arranging corresponding corpus participles according to the occurrence frequency from large to small, and extracting the corpus participles in the preset number and corresponding arrangement serial numbers to form a phrase list dictionary.
3. The method of claim 2, wherein the speech segmentation set comprises at least one row of first segmentation sequences formed after segmentation processing the speech of the user;
correspondingly, the forming of the voice input file of the user voice according to the voice word set and the pre-formed phrase list dictionary includes:
initializing and constructing a voice input matrix, wherein the number of rows of the voice input matrix is equal to the number of rows of a first participle sequence contained in the voice participle set, and the number of columns of the voice input matrix is equal to the number of participles of a first participle contained in the longest first participle sequence in the voice participle set;
aiming at each first participle in the voice participle set, if the first participle exists in the phrase list dictionary, taking a sequence number corresponding to the first participle as a target coding value; otherwise, using a preset coding value as a target coding value of the first word segmentation;
correspondingly writing target coding values of first participles contained in each participle sequence into the voice input matrix by taking the participle sequence as a row unit, and filling empty elements in the voice input matrix with the preset coding values;
and taking the assigned voice input matrix as a voice input file of the voice of the user.
4. The method of claim 2, wherein the sample corpus comprises at least one set of sample question-and-answer sentences, each set of sample question-and-answer sentences comprising a row of sample question-and-answer sentences and an associated row of sample answer-and-answer sentences;
accordingly, the process of determining the voice response template includes the steps of:
extracting sample question sentences and sample answer sentences in the sample corpus set to form a sample question sentence set and a sample answer sentence set;
forming a sample question file according to the sample question set, and forming a sample answer file and a sample proofreading file according to the sample answer set and the integrated statement expression rule;
determining standard semantic parameters corresponding to sample participles of each line of the sample question sentence file;
and combining each standard semantic parameter according to the sample answer file and the sample proofreading file to form a voice answer template comprising each standard semantic parameter and a corresponding standard answer participle sequence.
5. The method of claim 4, wherein forming a sample question file from the set of sample questions comprises:
performing word segmentation on the sample question set to obtain a question word set comprising at least one line of second word segmentation sequence;
initializing and constructing a sample question matrix, wherein the row number of the sample question matrix is equal to the row number of a second participle sequence contained in the question participle set, and the column number of the sample question matrix is equal to the participle number of a second participle contained in the longest second participle sequence in the question participle set;
aiming at each second participle in the sample question sentence set, taking the corresponding arrangement serial number of the second participle in the phrase list dictionary as a target coding value of the second participle;
correspondingly writing target coding values of second participles contained in each participle sequence into the sample question matrix by taking the participle sequence as a row unit, and filling empty elements in the sample question matrix with preset coding values;
and taking the sample question matrix after assignment as the sample question file.
6. The method of claim 4, wherein said forming a sample sentence file and a sample collation file according to said sample sentence set binding sentence expression rules comprises:
word segmentation processing is carried out on the sample answer sentence set, and an answer sentence word segmentation set comprising at least one line of third word segmentation sequence is obtained;
aiming at each third participle in the sample answer sentence set, taking the corresponding arrangement serial number of the third participle in the phrase list dictionary as a target coding value of the third participle;
initializing and constructing a sample answer sentence matrix, wherein the row number of the sample answer sentence matrix is equal to the row number of a third participle sequence contained in the answer sentence participle set, and the column number of the sample answer sentence matrix is equal to the participle number of a third participle contained in the longest third participle sequence in the answer sentence participle set;
correspondingly writing target coding values of third participles contained in each participle sequence into the sample answer sentence matrix by taking the participle sequence as a row unit, and filling null elements in the sample answer sentence matrix with preset coding values;
initializing and constructing a sample proofreading matrix, wherein the row number of the sample proofreading matrix is equal to the row number of a third participle sequence contained in the answer sentence participle set, and the column number of the sample proofreading matrix is equal to the participle number of a third participle contained in the longest third participle sequence in the answer sentence participle set;
correspondingly writing target coding values of third participles contained in each participle sequence into the sample proofreading matrix by taking the participle sequence as a row unit, and filling empty elements in the sample proofreading matrix with preset coding values;
and taking the sample sentence answering matrix after assignment as the sample sentence answering file, and taking the sample proofreading matrix after assignment as the sample proofreading file.
7. The method of claim 4, wherein combining each of the standard semantic parameters with the sample sentence file and the sample collation file to form a voice response template including each of the standard semantic parameters and a corresponding standard response participle sequence comprises:
matching the standard semantic parameters with the sample answer file to generate an initial voice answer template corresponding to each participle sequence in the answer file;
matching each participle sequence in the initial voice reply template with the sample proofreading file one by one, replacing participle sequences with different matching results with corresponding participle sequences in the sample proofreading file, taking the replaced participle sequences as the standard reply participle sequences, and generating the voice reply template with the standard semantic parameters corresponding to the standard reply participle sequences.
8. The method according to any one of claims 1-7, wherein said determining reply information corresponding to said user speech according to the search result comprises:
if the target semantic parameter is the standard semantic parameter in the voice reply template, taking the standard reply participle sequence corresponding to the target semantic parameter in the voice reply template as the reply information of the user voice; otherwise, the preset reply sentence is used as the reply information of the user voice.
9. A reply information determination device for a vehicle-mounted robot, comprising:
the voice acquisition module is used for performing word segmentation processing on the received user voice to obtain a voice word segmentation set of the user voice;
the file generation module is used for forming a voice input file of the user voice according to the voice word set and a pre-formed phrase list dictionary;
a template determination module for determining a voice response template based on a given sample corpus;
the parameter searching module is used for determining target semantic parameters corresponding to each row in the voice input file and searching the target semantic parameters in a predetermined voice reply template;
and the information determining module is used for determining the reply information corresponding to the user voice according to the search result.
10. An apparatus, characterized in that the apparatus comprises:
one or more processors;
storage means for storing one or more programs;
when the one or more programs are executed by the one or more processors, the one or more processors implement the reply information determination method of the in-vehicle robot according to any of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910990232.9A CN110956962A (en) | 2019-10-17 | 2019-10-17 | Reply information determination method, device and equipment for vehicle-mounted robot |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910990232.9A CN110956962A (en) | 2019-10-17 | 2019-10-17 | Reply information determination method, device and equipment for vehicle-mounted robot |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110956962A true CN110956962A (en) | 2020-04-03 |
Family
ID=69976400
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910990232.9A Pending CN110956962A (en) | 2019-10-17 | 2019-10-17 | Reply information determination method, device and equipment for vehicle-mounted robot |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110956962A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022046517A1 (en) * | 2020-08-28 | 2022-03-03 | Micron Technology, Inc. | Systems and methods for reducing latency in cloud services |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6728679B1 (en) * | 2000-10-30 | 2004-04-27 | Koninklijke Philips Electronics N.V. | Self-updating user interface/entertainment device that simulates personal interaction |
| CN106202349A (en) * | 2016-06-29 | 2016-12-07 | 杭州华三通信技术有限公司 | Web page classifying dictionary creation method and device |
| CN108399169A (en) * | 2017-02-06 | 2018-08-14 | 阿里巴巴集团控股有限公司 | Dialog process methods, devices and systems based on question answering system and mobile device |
| CN108491372A (en) * | 2018-01-31 | 2018-09-04 | 华南理工大学 | A kind of Chinese word cutting method based on seq2seq models |
| CN109492077A (en) * | 2018-09-29 | 2019-03-19 | 北明智通（北京）科技有限公司 | The petrochemical field answering method and system of knowledge based map |
| CN109522395A (en) * | 2018-10-12 | 2019-03-26 | 平安科技（深圳）有限公司 | Automatic question-answering method and device |
| CN109726396A (en) * | 2018-12-20 | 2019-05-07 | 泰康保险集团股份有限公司 | Semantic matching method, device, medium and the electronic equipment of question and answer text |
| CN109885810A (en) * | 2019-01-17 | 2019-06-14 | 平安城市建设科技(深圳)有限公司 | Nan-machine interrogation's method, apparatus, equipment and storage medium based on semanteme parsing |
| CN110046350A (en) * | 2019-04-12 | 2019-07-23 | 百度在线网络技术（北京）有限公司 | Grammatical bloopers recognition methods, device, computer equipment and storage medium |
| CN110109829A (en) * | 2019-04-15 | 2019-08-09 | 福建天晴在线互动科技有限公司 | Intelligent dialogue automates method of calibration, storage medium |
| CN110135551A (en) * | 2019-05-15 | 2019-08-16 | 西南交通大学 | A kind of robot chat method of word-based vector sum Recognition with Recurrent Neural Network |
| CN110209790A (en) * | 2019-06-06 | 2019-09-06 | 阿里巴巴集团控股有限公司 | Question and answer matching process and device |
-
2019
- 2019-10-17 CN CN201910990232.9A patent/CN110956962A/en active Pending
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6728679B1 (en) * | 2000-10-30 | 2004-04-27 | Koninklijke Philips Electronics N.V. | Self-updating user interface/entertainment device that simulates personal interaction |
| CN106202349A (en) * | 2016-06-29 | 2016-12-07 | 杭州华三通信技术有限公司 | Web page classifying dictionary creation method and device |
| CN108399169A (en) * | 2017-02-06 | 2018-08-14 | 阿里巴巴集团控股有限公司 | Dialog process methods, devices and systems based on question answering system and mobile device |
| CN108491372A (en) * | 2018-01-31 | 2018-09-04 | 华南理工大学 | A kind of Chinese word cutting method based on seq2seq models |
| CN109492077A (en) * | 2018-09-29 | 2019-03-19 | 北明智通（北京）科技有限公司 | The petrochemical field answering method and system of knowledge based map |
| CN109522395A (en) * | 2018-10-12 | 2019-03-26 | 平安科技（深圳）有限公司 | Automatic question-answering method and device |
| CN109726396A (en) * | 2018-12-20 | 2019-05-07 | 泰康保险集团股份有限公司 | Semantic matching method, device, medium and the electronic equipment of question and answer text |
| CN109885810A (en) * | 2019-01-17 | 2019-06-14 | 平安城市建设科技(深圳)有限公司 | Nan-machine interrogation's method, apparatus, equipment and storage medium based on semanteme parsing |
| CN110046350A (en) * | 2019-04-12 | 2019-07-23 | 百度在线网络技术（北京）有限公司 | Grammatical bloopers recognition methods, device, computer equipment and storage medium |
| CN110109829A (en) * | 2019-04-15 | 2019-08-09 | 福建天晴在线互动科技有限公司 | Intelligent dialogue automates method of calibration, storage medium |
| CN110135551A (en) * | 2019-05-15 | 2019-08-16 | 西南交通大学 | A kind of robot chat method of word-based vector sum Recognition with Recurrent Neural Network |
| CN110209790A (en) * | 2019-06-06 | 2019-09-06 | 阿里巴巴集团控股有限公司 | Question and answer matching process and device |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022046517A1 (en) * | 2020-08-28 | 2022-03-03 | Micron Technology, Inc. | Systems and methods for reducing latency in cloud services |
| US11817087B2 (en) | 2020-08-28 | 2023-11-14 | Micron Technology, Inc. | Systems and methods for reducing latency in cloud services |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110210029B (en) | Method, system, device and medium for correcting error of voice text based on vertical field | |
| CN108363790B (en) | Method, device, equipment and storage medium for evaluating comments | |
| CN108304375B (en) | Information identification method and equipment, storage medium and terminal thereof | |
| CN108959246B (en) | Answer selection method and device based on improved attention mechanism and electronic equipment | |
| CN108287858B (en) | Semantic extraction method and device for natural language | |
| CN108847241B (en) | Method for recognizing conference voice as text, electronic device and storage medium | |
| CN111444320B (en) | Text retrieval method and device, computer equipment and storage medium | |
| CN110377916B (en) | Word prediction method, word prediction device, computer equipment and storage medium | |
| CN110704621B (en) | Text processing method and device, storage medium and electronic equipment | |
| CN110737768B (en) | Text abstract automatic generation method and device based on deep learning and storage medium | |
| CN112100354B (en) | Man-machine conversation method, device, equipment and storage medium | |
| US20200301919A1 (en) | Method and system of mining information, electronic device and readable storable medium | |
| CN110276023B (en) | POI transition event discovery method, device, computing equipment and medium | |
| US11031009B2 (en) | Method for creating a knowledge base of components and their problems from short text utterances | |
| CN102439540A (en) | Input method editor | |
| CN112287684A (en) | Short text auditing method and device integrating variant word recognition | |
| CN110222328B (en) | Method, device and equipment for labeling participles and parts of speech based on neural network and storage medium | |
| CN112395385A (en) | Text generation method and device based on artificial intelligence, computer equipment and medium | |
| CN110032734B (en) | Training method and device for similar meaning word expansion and generation of confrontation network model | |
| EP3267327A1 (en) | Entailment pair expansion device, computer program therefor, and question-answering system | |
| CN103314369A (en) | Method and device for machine translation | |
| CN109522396B (en) | Knowledge processing method and system for national defense science and technology field | |
| CN111414746A (en) | Matching statement determination method, device, equipment and storage medium | |
| CN109710732A (en) | Information query method, device, storage medium and electronic equipment | |
| CN112185361A (en) | Speech recognition model training method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200403 |