Information processing method and device, electronic equipment, vehicle and storage medium

Abstract

The invention relates to an information processing method and device, electronic equipment, a vehicle and a storage medium. The method comprises the steps that target text information is acquired; extracting a first target phrase and a second target phrase which have an association relationship from the target text information by utilizing an identification model; processing the first target phrase, the second target phrase and an association relationship between the first target phrase and the second target phrase into a to-be-matched triple; in the target knowledge graph, determining at least one candidate triple matched with the to-be-matched triple; determining the similarity between each candidate triple and the triple to be matched; taking the candidate triad of which the similarity with the to-be-matched triad meets a preset condition as a target triad; and taking the object referred by the target triple as an identification result of the target text information. The method and the device can solve the problem that the description of the target object in the given target text information is inaccurate and the target object cannot be directly identified based on the target text information due to memory errors of a user.

Description

Technical Field

[0001] The present disclosure relates to the field of information processing technology, and in particular to an information processing method, device, electronic equipment, vehicle and storage medium. Background Art

[0002] With the continuous breakthroughs in artificial intelligence technology and the increasing popularity of various smart devices, human-computer interaction appears more and more frequently in people's daily work and life. In practice, users often hope to use human-computer interaction technology to control smart devices to perform operations corresponding to control instructions by issuing control instructions. In this process, accurate recognition of control instructions is required.

[0003] Taking the example of controlling a smart device to play multimedia resources by issuing voice commands, an existing information processing method is to pre-construct a multimedia resource name vocabulary; the words in the multimedia resource name vocabulary are standard names of multimedia resources, the control instruction is matched with the words in the multimedia resource name vocabulary, the multimedia resource referred to by the matching word is used as the multimedia resource to be played, and it is played. This method requires that the multimedia resource name in the control instruction issued by the user is the standard name of the multimedia resource. For example, assume that the standard name of a multimedia resource is "ABCD". If the user does not remember the standard name of the multimedia resource accurately, he will mistakenly remember it as "ABD" and issue the command "Play ABD". In this case, because "ABD" and "ABCD" are different, the two cannot be matched successfully, resulting in the failure to play the multimedia resource corresponding to the control instruction. Summary of the invention

[0004] In order to solve the above technical problems, the present disclosure provides an information processing method, device, electronic device, vehicle and storage medium.

[0005] In a first aspect, the present disclosure provides an information processing method, comprising:

[0006] Get target text information;

[0007] Extracting a first target phrase and a second target phrase having an associated relationship from the target text information using a recognition model;

[0008] Processing the first target phrase, the second target phrase, and the association relationship between the first target phrase and the second target phrase into a triple to be matched;

[0009] In the target knowledge graph, determining at least one candidate triple that matches the triple to be matched;

[0010] Determine the similarity between each of the candidate triples and the triple to be matched;

[0011] Taking the candidate triplet whose similarity with the triplet to be matched meets a preset condition as the target triplet;

[0012] The object referred to by the target triplet is used as the recognition result of the target text information.

[0013] In a second aspect, the present disclosure further provides an information processing device, including:

[0014] An acquisition module is used to acquire target text information;

[0015] An extraction module, used for extracting a first target phrase and a second target phrase having an associated relationship from the target text information by using a recognition model;

[0016] an integration module, configured to process the first target phrase, the second target phrase, and the association relationship between the first target phrase and the second target phrase into a triple to be matched;

[0017] A matching module, used to determine at least one candidate triple matching the triple to be matched in the target knowledge graph;

[0018] A calculation module, used to determine the similarity between each of the candidate triples and the triple to be matched;

[0019] A target determination module, configured to take the candidate triplet whose similarity with the triplet to be matched meets a preset condition as a target triplet;

[0020] The recognition module is used to use the object referred to by the target triple as the recognition result of the target text information.

[0021] In a third aspect, the present disclosure further provides an electronic device, including: a processor and a memory;

[0022] The processor is used to execute the steps of any of the above methods by calling the program or instruction stored in the memory.

[0023] In a fourth aspect, the present disclosure further provides a vehicle, comprising any of the above-mentioned devices or any of the above-mentioned electronic devices.

[0024] In a fifth aspect, the present disclosure further provides a computer-readable storage medium, which stores a program or instruction, and the program or instruction enables a computer to execute the steps of any of the above methods.

[0025] Compared with the prior art, the technical solution provided by the embodiments of the present disclosure has the following advantages:

[0026] The technical solution provided by the embodiment of the present disclosure is to extract the first target phrase and the second target phrase with an associated relationship from the target text information by setting up a recognition model; process the first target phrase, the second target phrase and the associated relationship between the first target phrase and the second target phrase as a triple to be matched; determine at least one candidate triple that matches the triple to be matched in the target knowledge graph; determine the similarity between each candidate triple and the triple to be matched; use the candidate triple whose similarity with the triple to be matched meets the preset conditions as the target triple; and use the object referred to by the target triple as the recognition result of the target text information. It can solve the problem that the description of the target object in the given target text information is inaccurate due to user memory errors, resulting in the inability to directly identify the target object based on the target text information, and can improve the accuracy of the recognition result of the target text information. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0028] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0029] Figure 1 is a flow chart of an information processing method in an embodiment of the present disclosure;

[0030] Figure 2 A flowchart of a method for training a recognition model provided in an embodiment of the present disclosure;

[0031] Figure 3 A flowchart of a method for implementing inputting an original sample sentence into a recognition model to obtain a first reference phrase and a second reference phrase provided in an embodiment of the present disclosure;

[0032] Figure 4 A schematic diagram of a sentence tree provided by an embodiment of the present disclosure;

[0033] Figure 5 is a flowchart of another information processing method in an embodiment of the present disclosure;

[0034] Figure 6 is a schematic diagram of the structure of an information processing device in an embodiment of the present disclosure;

[0035] Figure 7It is a structural schematic diagram of an electronic device in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0036] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0037] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0038] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0039] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0040] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0041] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0042] Figure 1This is a flowchart of an information processing method in an embodiment of the present disclosure. This embodiment is applicable to the case where information processing is performed in a client. The method can be executed by an information processing device, which can be implemented in software and / or hardware. The device can be configured in an electronic device, such as a terminal, specifically including but not limited to smart phones, PDAs, tablet computers, wearable devices with display screens, desktop computers, laptop computers, all-in-one machines, smart home devices, and vehicles. Alternatively, this embodiment is applicable to the case where information processing is performed in a server. The method can be executed by an information processing device, which can be implemented in software and / or hardware. The device can be configured in an electronic device, such as a server.

[0043] like Figure 1 As shown, the method may specifically include:

[0044] S110: Obtain target text information.

[0045] The target text information is information in text form that needs to be processed or recognized. Exemplarily, the target text information is a control instruction in text form.

[0046] In one embodiment, the terminal may be configured to monitor the user's voice information, and when the voice information input by the user is monitored, the voice information is converted into target text information.

[0047] Furthermore, an automatic speech recognition (ASR) engine may be used to convert speech information into target text information.

[0048] S120 , extracting a first target phrase and a second target phrase having an associated relationship from the target text information using a recognition model.

[0049] The target text information is essentially a sentence, which is composed of a plurality of phrases. The first target phrase and the second target phrase are both phrases that constitute the target text information.

[0050] In the target text information, the first target phrase and the second target phrase are used to jointly define the same target object. Optionally, the first target phrase and the second target phrase may be phrases that describe the target object. For example, the first target phrase and the second target phrase may be phrases that describe the properties of the target object. The target object is the specific thing that the user intends to point to. Exemplarily, in a scenario where the in-vehicle device is controlled to play multimedia through the target text information, the target object is the multimedia resource that the user intends to play.

[0051] In the present application, the specific relationship between the first target phrase and the second target phrase is not limited in the present application.

[0052] Exemplarily, in a scenario where the target text information is an instruction for controlling the vehicle-mounted device to play multimedia, of the first target phrase and the second target phrase, one is a phrase that can be used as a multimedia identifier, and the other is a phrase that can be used as multimedia auxiliary information. For example, the target text information is "Play XX's ABCD", where "XX" is the name of the singer and "ABCD" is the name of the song. When executing this step, "XX" can be used as the first target phrase and "ABCD" can be used as the second target phrase. In practice, there are multiple singers who have sung a song named "ABCD", that is, there are multiple songs named "ABCD", and the differences between these songs include different singers. Through the first target phrase and the second target phrase, it can be clear which song the user wants to play, and the song that the user wants to play is the target object.

[0053] It should be noted that due to the user's memory errors, the first target phrase and the second target phrase may not accurately describe the target object. This situation is specifically manifested in that the first target phrase and the second target phrase are similar words of the standard phrase. For example, if the standard name of a multimedia resource is "ABCEDDD", the user may mistakenly remember the standard name as "ABCEDD", "ABECDDD" or "MMM", etc. "ABCEDD", "ABECDDD" or "MMM" are all similar to "ABCEDDD". Some characters of "ABCEDD" and "ABCEDDD" are the same. The characters of "ABCEDD" and "ABECDDD" are the same, but the order of character arrangement is different. "MMM" and "ABECDDD" have the same semantics.

[0054] In this step, the recognition model is a trained recognition model, which can extract the first target phrase and the second target phrase that may be used to define the target object from the target text information.

[0055] S130: Process the first target phrase, the second target phrase, and the association relationship between the first target phrase and the second target phrase into triples to be matched.

[0056] Processing the first target phrase, the second target phrase, and the association relationship between the first target phrase and the second target phrase as triples to be matched means treating the first target phrase, the second target phrase, and the association relationship between the first target phrase and the second target phrase as a whole.

[0057] S140. In the target knowledge graph, determine at least one candidate triple that matches the triple to be matched.

[0058] In this step, the target knowledge graph is a pre-specified knowledge graph. In practice, the information processing method provided by this application can be applied to any field. The target knowledge graph is a knowledge graph of a pre-specified field. Exemplarily, in a scenario where a vehicle-mounted device is controlled to play multimedia through target text information, the target knowledge graph is a multimedia knowledge graph.

[0059] The target knowledge graph can be viewed as a data set consisting of many triples. In the target knowledge graph, any triple consists of two nodes and an edge connecting the two nodes. A triple uniquely points to an object. The phrase as a node in the triple is used to describe a certain aspect of the attribute of the object pointed to by the triple, and the phrase as a node is an accurate term to describe the attribute of the object.

[0060] The purpose of introducing the target knowledge graph here is that when a user request has a contextual relationship (i.e., an association relationship), a rich graph relationship network can be used to assist decision-making and determine the user's intention.

[0061] S150: Determine the similarity between each candidate triple and the triple to be matched.

[0062] There are many methods for implementing this step, and this application does not limit this. Exemplarily, determine the vectors of each candidate triple and the vectors of the triples to be matched; calculate the cosine value of the angle between the vector of the triple to be matched and the vectors of each candidate triple; based on the cosine value of the angle between the vector of the triple to be matched and the vectors of each candidate triple, determine the similarity between each candidate triple and the triple to be matched; or, determine the edit distance between each candidate triple and the triple to be matched; based on the edit distance between each candidate triple and the triple to be matched, determine the similarity between each candidate triple and the triple to be matched.

[0063] It should be noted that when determining the similarity between the candidate triples and the triples to be matched, the association between the two nodes of each triple and the two nodes is considered, so that the similarity between the candidate triples and the triples to be matched can be calculated accurately.

[0064] S160: taking a candidate triplet whose similarity with the triplet to be matched meets a preset condition as a target triplet.

[0065] The preset condition is a criterion for measuring whether to use a candidate triple as a target triple. The present application does not limit the content included in it. Exemplarily, the preset condition includes that the similarity with the triple to be matched is greater than a set threshold.

[0066] S170: Taking the object referred to by the target triple as the recognition result of the target text information.

[0067] The essence of this step is to determine the object referred to by the target triple as the object that the user intends to point to (ie, the target object).

[0068] The above technical solution is set to use the recognition model to extract the first target phrase and the second target phrase with an associated relationship from the target text information; the first target phrase, the second target phrase and the associated relationship between the first target phrase and the second target phrase are processed as a to-be-matched triple; in the target knowledge graph, at least one candidate triple matching the to-be-matched triple is determined; the similarity between each candidate triple and the to-be-matched triple is determined; the candidate triple whose similarity with the to-be-matched triple meets the preset conditions is used as the target triple; the object referred to by the target triple is used as the recognition result of the target text information. Its essence is to first identify the to-be-matched triple in the target text information that may be used to describe the target object, and then determine the target triple with a high similarity to the to-be-matched triple from the target knowledge graph, and determine the object pointed to by the user's intention through the target triple. It can solve the problem that the description of the target object in the given target text information is inaccurate due to the user's memory error, resulting in the inability to directly identify the target object based on the target text information, and can improve the accuracy of the target text information recognition result.

[0069] Those skilled in the art will appreciate that, before the above technical method is executed, the recognition model needs to be trained so that the recognition model has the function of extracting the first target phrase and the second target phrase having an associated relationship from the target text information.

[0070] There are many methods for training the recognition model, and this application does not limit this. Figure 2 A flowchart of a method for training a recognition model provided by an embodiment of the present disclosure. Figure 2 , the method comprising:

[0071] S210 , obtaining sample data, where the sample data includes an original sample sentence and annotation information corresponding to the original sample sentence, where the annotation information is used to indicate a first sample phrase and a second sample phrase that have an associated relationship in the original sample sentence.

[0072] Optionally, in practice, the original sample sentences can be collected by crawler technology, and the collected original sample sentences can be annotated using the BIOES (B-begin, I-inside, O-outside, E-end, S-single) method to obtain the annotation information corresponding to the original sample sentences. In the BIOES method, B-begin represents the beginning of the entity; I-inside represents the middle of the entity; O-outside represents a non-entity, which is used to mark irrelevant characters; E-end represents the end of the entity; S-single represents a single character, which is an entity itself. For example, an original sample sentence is "I want to watch XXX's YYY". Among them, "XXX" represents the name of the actor. "YYY" represents the standard name of the film and television drama. The result of the annotation of the original sample sentence by the BIOES method is "OOOB-ACTORS I-ACTORS E-ACTORS OB-TITLEI-TITLE E-TITLE".

[0073] S220: Use sample data to train the recognition model.

[0074] There are many specific implementation methods for this step, and this application does not limit this. Exemplarily, the implementation method of this step includes: inputting the original sample sentence into the recognition model to obtain a first reference phrase and a second reference phrase with an associated relationship; processing the first reference phrase, the second reference phrase, and the associated relationship between the first reference phrase and the second reference phrase into a first triple; processing the first sample phrase, the second sample phrase, and the associated relationship between the first sample phrase and the second sample phrase into a second triple; determining a reward score based on the similarity between the first triple and the second triple; and modifying the parameters of the recognition model with the goal of obtaining the maximum reward score.

[0075] The reward score is an indicator to measure the performance of the recognition model. The higher the similarity between the first triplet and the second triplet, the higher the reward score, which means the current recognition model has a better performance. The lower the similarity between the first triplet and the second triplet, the lower the reward score, which means the current recognition model has a poor performance.

[0076] The essence of this setting is to use reinforcement learning methods to train the recognition model. The recognition model obtains the maximum reward score by selecting the correct reference phrase, which can make the recognition result of the finally trained recognition model more accurate, laying the foundation for the final accurate recognition of the target text information.

[0077] Further, see Figure 3 The method of inputting the original sample sentence into the recognition model to obtain the first reference phrase and the second reference phrase may include:

[0078] S221. Segment the original sample sentence to obtain multiple candidate sample phrases.

[0079] There are many methods for implementing this step, and this application does not limit this. Exemplarily, the target text information can be segmented using the Jieba word segmentation algorithm or other word segmentation algorithms.

[0080] S222. Expand at least part of the multiple candidate sample phrases using the target knowledge graph to obtain a sentence tree.

[0081] There are many ways to implement this step, and this application does not limit this. Exemplarily, the specific implementation method of this step includes: using each candidate phrase as an index in turn, searching for an extended phrase corresponding to the candidate phrase in the target knowledge graph; and concatenating each extended phrase with the target text information to obtain a sentence tree.

[0082] Exemplarily, the target text information is "play XX's ABCD", where "XX" is the name of the singer and "ABCD" is the name of the song. "The candidate phrases obtained by segmenting the target text information include "play", "XX", "of", and "ABCD". "Play", "XX", "of", and "ABCD" are respectively used as indexes to query in the multimedia knowledge graph. Assume that there are no triples corresponding to "play" and "of" after the query, and the expansion of "play" and "of" fails. A triple corresponding to "XX" is found as "XX-performance-EFZG", where "EFZG" is the name of the TV series. "Performance" and "EFZG" are used as the expansion words corresponding to "XX". A triple corresponding to "ABCD" is found as "YYY-lyrics-ABCD", where "YYY" is a person's name. "Lyrics" and "YYY" are used as the expansion words corresponding to "ABCD". "Performance", "EFZG", "lyrics", and "YYY" are concatenated with the target text information to obtain a sentence tree. The sentence tree is as follows: Figure 4 shown.

[0083] S223, determining the position encoding result vector of the sentence tree and the visibility of any two phrases in the sentence tree.

[0084] There are many methods for implementing this step, and this application does not limit this. Exemplarily, the specific implementation method of this step includes: inserting the extended word into the sentence tree, flattening it into a sequence, position encoding the sequence, and obtaining a position encoding result vector; according to whether any two phrases in the sentence tree are in the same branch, determining the visibility of the two phrases in the sentence tree. The visibility of any two phrases can be represented by a visibility matrix.

[0085] S224. Input the position encoding result vector of the sentence tree and the visibility of any two phrases in the sentence tree into the recognition model to obtain a first reference phrase and a second reference phrase having an associated relationship.

[0086] By setting the original sample sentence to be segmented, multiple candidate sample phrases are obtained; at least part of the multiple candidate sample phrases are expanded using the target knowledge graph to obtain a sentence tree, the purpose of which is to inject the association relationship information of the two phrases into the sentence with the help of the knowledge graph, so that the recognition model can learn and absorb professional domain knowledge. Setting the position encoding result vector of the sentence tree and the visibility of any two phrases in the sentence tree is determined; the position encoding result vector of the sentence tree and the visibility of any two phrases in the sentence tree are input into the recognition model to obtain a first reference phrase and a second reference phrase with an association relationship, the purpose of which is to suppress the occurrence of an undesirable situation in which the sentence deviates from its correct meaning due to excessive consideration of domain knowledge, thereby improving the recognition accuracy of the trained recognition model.

[0087] On the basis of the above technical solutions, optionally, in S223, the determined position encoding result vector of the sentence tree may be a soft position (Soft-position) encoding result vector of the sentence tree, or a hard position (Hard-position) encoding result vector.

[0088] In another embodiment, determining the position coding result vector of the sentence tree may include: determining a soft position coding result vector and a hard position coding result vector of the sentence tree; determining a final position coding result vector of the sentence tree based on the soft position coding result vector and the hard position coding result vector of the sentence tree; S224 may be replaced by: inputting the final position coding result vector of the sentence tree and the visibility of any two phrases in the sentence tree into the recognition model to obtain a first reference phrase and a second reference phrase with an associated relationship.

[0089] Among them, there are many specific implementation methods of "determining the final position coding result vector of the sentence tree based on the soft position coding result vector and the hard position coding result vector of the sentence tree", and the present application does not limit this. Exemplarily, the specific implementation method of "determining the final position coding result vector of the sentence tree based on the soft position coding result vector and the hard position coding result vector of the sentence tree" includes: averaging the soft position coding result vector and the hard position coding result vector of the sentence tree, and taking the average value as the final position coding result vector of the sentence tree. This setting can make the final position coding result vector more accurate in describing the structure of the sentence tree, which is conducive to shortening the time spent on training the recognition model, reducing the recognition error of the recognition model, and improving the recognition accuracy of the recognition model.

[0090] In the above technical solution, the target knowledge graph is used in the process of training the recognition model and in the process of determining at least one candidate triple that matches the triple to be matched. In practice, there are many methods for constructing the target knowledge graph, and this application does not limit this. Optionally, the method for constructing the target knowledge graph includes: obtaining a first phrase; querying an encyclopedia database to obtain a second phrase associated with the first phrase; using the first phrase and the second phrase as nodes and the association relationship between the first phrase and the second phrase as an edge to form a triple.

[0091] The first phrase is a pre-given phrase. The second phrase is a phrase expanded with the help of an encyclopedia database. The first phrase and the second phrase are associated. The application does not limit the specific relationship between the first phrase and the second phrase. The first phrase and the second phrase may describe the same object.

[0092] This setting can realize the automatic construction of the target knowledge graph.

[0093] Exemplarily, if the target knowledge graph is a multimedia knowledge graph, the method for constructing the multimedia knowledge graph includes:

[0094] The method comprises the following steps: extracting information from multimedia resources associated with an audio and video website to obtain multimedia standard identification information; using the multimedia standard identification information as a first phrase; querying an encyclopedia database to obtain auxiliary information associated with each multimedia standard identification information, and determining the auxiliary information as a second phrase; and forming a triplet by using the first phrase and the second phrase as nodes and the association relationship between the first phrase and the second phrase as an edge.

[0095] Optionally, the multimedia standard identification information includes a standard name or alias of the multimedia.

[0096] The auxiliary information may be a phrase that describes the attributes of the multimedia resource referred to by the first phrase. The attributes of multimedia resources refer to the properties of multimedia resources during the process of multimedia resource creation, recording, generation or dissemination. Exemplarily, if the multimedia resource is music, the auxiliary information may include at least one of the following: singer, lyricist, composer, album, music style, release date, song language, creative background, lyrics and award record; if the multimedia resource is a film or television work, the auxiliary information may include at least one of the following: type, director, actor, screenwriter, plot summary, role, award record, production time and dialogue voice.

[0097] By setting up information extraction for multimedia resources associated with audio and video websites, multimedia standard identification information is obtained; using the multimedia standard identification information as the first phrase can ensure that the constructed multimedia knowledge graph can cover the playable multimedia resources in the audio and video websites as much as possible.

[0098] By setting a query encyclopedia database to obtain auxiliary information associated with each multimedia standard identification information, and determining the auxiliary information as the second phrase, it can be ensured that the obtained second phrase is an accurate description of the multimedia resource attribute.

[0099] In existing information processing methods, in order to improve the recognition accuracy of multimedia resource playback instructions, it is usually necessary to expand the multimedia resource standard name or alias to obtain similar words of the multimedia resource standard name or alias, and add the similar words of the multimedia resource standard name or alias to the multimedia resource name vocabulary. In practice, when this method is used to expand the multimedia resource standard name or alias, the completeness of the multimedia resource name vocabulary is often poor due to the large number of expansion directions. However, when the technical solution provided by the present application is used, when constructing a multimedia knowledge graph, it is not necessary to expand the multimedia resource standard name or alias, which can reduce the difficulty of constructing the multimedia knowledge graph.

[0100] Furthermore, if the target knowledge graph is a multimedia knowledge graph, after forming a triple with the first phrase and the second phrase as nodes and the association between the first phrase and the second phrase as an edge, it can also include: associating the multimedia resources associated with the audio and video website with the triple.

[0101] Associating the multimedia resources associated with the audio and video website with the triplet may include: establishing a corresponding relationship between the playback address of the multimedia resources in the audio and video website and the triplet, and storing the corresponding relationship.

[0102] By associating the multimedia resources associated with the audio and video website with the triples, each triple in the multimedia knowledge graph can uniquely point to the playback address of a multimedia resource, making it easy to quickly find the playable multimedia resources it points to based on the target triple after a certain triple is determined as the target triple.

[0103] Figure 5 A flowchart of another information processing method provided by an embodiment of the present disclosure. Figure 5 , the information processing method comprises:

[0104] S310, obtaining a voice control instruction; the voice control instruction is used to control the electronic device to play multimedia resources.

[0105] S320: Convert the voice control instruction into target text information.

[0106] S330: Extracting a first target phrase and a second target phrase having an associated relationship from the target text information using a recognition model.

[0107] S340: Process the first target phrase and the second target phrase into triples to be matched.

[0108] S350. In the target knowledge graph, determine at least one candidate triple that matches the triple to be matched.

[0109] S360: Determine the similarity between each candidate triple and the triple to be matched.

[0110] S370: Take the candidate triplet whose similarity with the triplet to be matched meets the preset condition as the target triplet.

[0111] S380: Use the multimedia resources associated with the target triple as the recognition result of the target text information.

[0112] S390: Play the recognition result.

[0113] The above technical solution can solve the problem that the description information of multimedia resources in the multimedia resource playback instruction given is inaccurate due to user memory errors, and it is impossible to identify which multimedia resource the user wants to play. It can improve the accuracy of multimedia resource determination in multimedia resource playback instructions. And because the multimedia knowledge graph that has been constructed before includes the association relationship between multimedia resources and triples associated with audio and video websites, the playback address of multimedia resources can be directly obtained through the multimedia knowledge graph. It can improve the efficiency of searching and playing multimedia resources.

[0114] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0115] Figure 6 The information processing device provided in the embodiment of the present disclosure may be configured in a client or in a server, and the information processing device specifically includes:

[0116] An acquisition module 610 is used to acquire target text information;

[0117] An extraction module 620 is used to extract a first target phrase and a second target phrase having an associated relationship from the target text information by using a recognition model;

[0118] An integration module 630, configured to process the first target phrase, the second target phrase, and the association relationship between the first target phrase and the second target phrase into a triple to be matched;

[0119] A matching module 640 is used to determine at least one candidate triple that matches the triple to be matched in the target knowledge graph;

[0120] A calculation module 650 is used to determine the similarity between each of the candidate triples and the triple to be matched;

[0121] A target determination module 660 is used to take the candidate triplet whose similarity with the triplet to be matched meets a preset condition as a target triplet;

[0122] The recognition module 670 is configured to use the object referred to by the target triple as the recognition result of the target text information.

[0123] Furthermore, the device also includes a model training module, which is used to:

[0124] Acquire sample data, the sample data including an original sample sentence and annotation information corresponding to the original sample sentence, the annotation information being used to indicate a first sample phrase and a second sample phrase having an associated relationship in the original sample sentence;

[0125] Inputting the original sample sentence into the recognition model to obtain a first reference phrase and a second reference phrase having an associated relationship;

[0126] Processing the first reference word group, the second reference word group, and the association relationship between the first reference word group and the second reference word group into a first triple;

[0127] Processing the first sample phrase, the second sample phrase, and the association relationship between the first sample phrase and the second sample phrase into a second triple;

[0128] Determining a reward score based on the similarity between the first triple and the second triple;

[0129] With the goal of obtaining the maximum reward score, the parameters of the recognition model are modified.

[0130] Furthermore, the model training module is used to:

[0131] Segmenting the original sample sentence to obtain multiple candidate sample phrases;

[0132] Expanding at least part of the plurality of candidate sample phrases using the target knowledge graph to obtain a sentence tree;

[0133] Determine the position encoding result vector of the sentence tree and the visibility of any two phrases in the sentence tree;

[0134] The position encoding result vector of the sentence tree and the visibility of any two phrases in the sentence tree are input into the recognition model to obtain a first reference phrase and a second reference phrase having an associated relationship.

[0135] Furthermore, the model training module is used to:

[0136] Determine a soft position encoding result vector and a hard position encoding result vector of the sentence tree;

[0137] Determine a final position encoding result vector of the sentence tree based on the soft position encoding result vector and the hard position encoding result vector of the sentence tree;

[0138] The final position encoding result vector of the sentence tree and the visibility of any two phrases in the sentence tree are input into the recognition model to obtain a first reference phrase and a second reference phrase with an associated relationship.

[0139] Furthermore, the device also includes a graph construction module, which is used to:

[0140] Get the first phrase;

[0141] Querying an encyclopedia database to obtain a second phrase associated with the first phrase;

[0142] A triple is formed by taking the first word group and the second word group as nodes and taking the association relationship between the first word group and the second word group as an edge.

[0143] Furthermore, the target knowledge graph is a multimedia knowledge graph, and the graph construction module obtains the first phrase, including:

[0144] Extracting information from multimedia resources associated with the audio and video website to obtain multimedia standard identification information; using the multimedia standard identification information as the first phrase;

[0145] The graph construction module queries the encyclopedia database to obtain a second phrase associated with the first phrase; including: querying the encyclopedia database to obtain auxiliary information associated with each of the multimedia standard identification information, and determining the auxiliary information as the second phrase;

[0146] The graph construction module uses the first phrase and the second phrase as nodes and the association relationship between the first phrase and the second phrase as an edge. After forming a triple, the module includes: associating the multimedia resources associated with the audio and video website with the triple.

[0147] The information processing device provided in the embodiment of the present disclosure can execute the steps executed by the client or server in the information processing method provided in the embodiment of the method of the present disclosure, and the execution steps and beneficial effects are no longer repeated here.

[0148] Figure 7 Schematic diagram of the structure of an electronic device in the embodiment of the present disclosure. Figure 7 , which shows a schematic diagram of the structure of an electronic device 500 suitable for implementing the embodiment of the present disclosure. The electronic device 500 in the embodiment of the present disclosure may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle terminals (such as vehicle navigation terminals), wearable electronic devices, etc., and fixed terminals such as digital TVs, desktop computers, smart home devices, etc. Figure 7 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0149] like Figure 7 As shown, the electronic device 500 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 to a random access memory (RAM) 503 to implement the information processing method of the embodiment described in the present disclosure. In the RAM 503, various programs and data required for the operation of the electronic device 500 are also stored. The processing device 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0150] Typically, the following devices may be connected to the I / O interface 505: an input device 506 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 507 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 508 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 509. The communication device 509 may allow the electronic device 500 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 7 The electronic device 500 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0151] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains a program code for executing the method shown in the flowchart, thereby implementing the information processing method as described above. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 509, or installed from the storage device 508, or installed from the ROM 502. When the computer program is executed by the processing device 501, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.

[0152] It should be noted that the computer-readable medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0153] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0154] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0155] The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device:

[0156] Get target text information;

[0157] Extracting a first target phrase and a second target phrase having an associated relationship from the target text information using a recognition model;

[0158] Processing the first target phrase, the second target phrase, and the association relationship between the first target phrase and the second target phrase into a triple to be matched;

[0159] In the target knowledge graph, determining at least one candidate triple that matches the triple to be matched;

[0160] Determine the similarity between each of the candidate triples and the triple to be matched;

[0161] Taking the candidate triplet whose similarity with the triplet to be matched meets a preset condition as the target triplet;

[0162] The object referred to by the target triplet is used as the recognition result of the target text information.

[0163] Optionally, when the above one or more programs are executed by the electronic device, the electronic device may also execute other steps described in the above embodiments.

[0164] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0165] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0166] The units involved in the embodiments described in the present disclosure may be implemented by software or hardware, wherein the name of a unit does not, in some cases, limit the unit itself.

[0167] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0168] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0169] According to one or more embodiments of the present disclosure, the present disclosure provides an electronic device, including:

[0170] one or more processors;

[0171] A memory for storing one or more programs;

[0172] When the one or more programs are executed by the one or more processors, the one or more processors implement any information processing method provided in the present disclosure.

[0173] According to one or more embodiments of the present disclosure, the present disclosure provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any information processing method provided by the present disclosure.

[0174] An embodiment of the present disclosure further provides a vehicle, which includes any information processing device provided by the present disclosure or any electronic device provided by the present disclosure.

[0175] Since the vehicle provided by the embodiment of the present disclosure includes the information processing device or electronic device provided by the present disclosure, it has the same or corresponding beneficial effects as the information processing device or electronic device included therein, which will not be described in detail here.

[0176] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other to form a technical solution.

[0177] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0178] Although the subject matter has been described in language specific to structural features and / or methodological logical actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims.

Claims

1. An information processing method, characterized in that: include: Get target text information; Extracting a first target phrase and a second target phrase having an associated relationship from the target text information using a recognition model; Processing the first target phrase, the second target phrase, and the association relationship between the first target phrase and the second target phrase into a triple to be matched; In the target knowledge graph, determining at least one candidate triple that matches the triple to be matched; Determine the similarity between each of the candidate triples and the triple to be matched; Taking the candidate triplet whose similarity with the triplet to be matched meets a preset condition as the target triplet; The object referred to by the target triplet is used as the recognition result of the target text information.

2. The method according to claim 1, characterized in that The method for training the recognition model includes: Acquire sample data, the sample data including an original sample sentence and annotation information corresponding to the original sample sentence, the annotation information being used to indicate a first sample phrase and a second sample phrase having an associated relationship in the original sample sentence; Inputting the original sample sentence into the recognition model to obtain a first reference phrase and a second reference phrase having an associated relationship; Processing the first reference word group, the second reference word group, and the association relationship between the first reference word group and the second reference word group into a first triple; Processing the first sample phrase, the second sample phrase, and the association relationship between the first sample phrase and the second sample phrase into a second triple; Determining a reward score based on the similarity between the first triple and the second triple; With the goal of obtaining the maximum reward score, the parameters of the recognition model are modified.

3. The method according to claim 2, characterized in that The step of inputting the original sample sentence into the recognition model to obtain a first reference phrase and a second reference phrase having an associated relationship includes: Segmenting the original sample sentence to obtain multiple candidate sample phrases; Expanding at least part of the plurality of candidate sample phrases using the target knowledge graph to obtain a sentence tree; Determine the position encoding result vector of the sentence tree and the visibility of any two phrases in the sentence tree; The position encoding result vector of the sentence tree and the visibility of any two phrases in the sentence tree are input into the recognition model to obtain a first reference phrase and a second reference phrase having an associated relationship.

4. The method according to claim 3, characterized in that The step of determining the position encoding result vector of each phrase in the sentence tree comprises: Determine a soft position encoding result vector and a hard position encoding result vector of the sentence tree; Determine a final position encoding result vector of the sentence tree based on the soft position encoding result vector and the hard position encoding result vector of the sentence tree; The step of inputting the position encoding result vector of the sentence tree and the visibility of any two phrases in the sentence tree into the recognition model to obtain a first reference phrase and a second reference phrase having an associated relationship comprises: The final position encoding result vector of the sentence tree and the visibility of any two phrases in the sentence tree are input into the recognition model to obtain a first reference phrase and a second reference phrase with an associated relationship.

5. The method according to claim 1, characterized in that Methods for constructing the target knowledge graph include: Get the first phrase; Querying an encyclopedia database to obtain a second phrase associated with the first phrase; A triple is formed by taking the first word group and the second word group as nodes and taking the association relationship between the first word group and the second word group as an edge.

6. The method according to claim 5, characterized in that The target knowledge graph is a multimedia knowledge graph, The obtaining of the first phrase includes: Extracting information from multimedia resources associated with the audio and video website to obtain multimedia standard identification information; using the multimedia standard identification information as the first phrase; The querying of the encyclopedia database to obtain a second phrase associated with the first phrase comprises: querying an encyclopedia database to obtain auxiliary information associated with each of the multimedia standard identification information, and determining the auxiliary information as a second phrase; The method of forming a triple with the first word group and the second word group as nodes and the association between the first word group and the second word group as an edge includes: The multimedia resources associated with the audio and video website are associated with the triplet.

7. An information processing device, characterized in that: include: An acquisition module is used to acquire target text information; An extraction module, used for extracting a first target phrase and a second target phrase having an associated relationship from the target text information by using a recognition model; an integration module, configured to process the first target phrase, the second target phrase, and the association relationship between the first target phrase and the second target phrase into a triple to be matched; A matching module, used to determine at least one candidate triple matching the triple to be matched in the target knowledge graph; A calculation module, used to determine the similarity between each of the candidate triples and the triple to be matched; A target determination module, configured to take the candidate triplet whose similarity with the triplet to be matched meets a preset condition as a target triplet; The recognition module is used to use the object referred to by the target triple as the recognition result of the target text information.

8. An electronic device, characterized in that: include: Processor and memory; The processor is used to execute the steps of the method according to any one of claims 1 to 6 by calling the program or instruction stored in the memory.

9. A vehicle, characterized in that: Comprising the apparatus as claimed in claim 7 or the electronic device as claimed in claim 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program or instruction, and the program or instruction enables a computer to execute the steps of the method according to any one of claims 1 to 6.

Images