WO2023206703A1

WO2023206703A1 - Event slot extraction method and apparatus, storage medium and electronic apparatus

Info

Publication number: WO2023206703A1
Application number: PCT/CN2022/096436
Authority: WO
Inventors: 苑春明
Original assignee: 青岛海尔科技有限公司; 海尔智家股份有限公司
Priority date: 2022-04-29
Filing date: 2022-05-31
Publication date: 2023-11-02
Also published as: CN117010364A

Abstract

An event slot extraction method and apparatus, a storage medium and an electronic apparatus, relating to the technical field of smart home/home automation, the method comprising: acquiring a target control statement of an event slot to be extracted, the target control statement comprising a plurality of words; performing syntactic analysis on the target control statement, and determining a group of candidate words from among the plurality of words, the number of the candidate words included in the group of candidate words being smaller than the number of words included in the plurality of words; according to the syntactic relationship between each candidate word in the group of candidate words and an associated word of said candidate word, determining a target word from among the group of candidate words, a specified syntactic relationship being between the target word and the associated word of the target word; and determining the target word and the associated word of the target word as a target event slot of the target control statement.

Description

Event slot extraction method and device, storage medium and electronic device

This disclosure claims the priority of the Chinese patent application submitted to the China Patent Office on April 29, 2022, with application number 202210468880.X and invention title "Event Slot Extraction Method and Device, Storage Medium and Electronic Device", all of which The contents are incorporated by reference into this disclosure.

Technical field

The present disclosure relates to the field of communications, and specifically, to an event slot extraction method and device, a storage medium, and an electronic device.

Background technique

In the field of intelligent question answering, user intentions can be understood through slot extraction. Taking slot extraction in the schedule field in the question and answer system as an example, the slots involved can include slot information such as time, events, singers, and songs. Compared with other slots, event slots have richer content, which makes their extraction more complex.

In the related art, the dictionary matching method is usually used to extract the event slot from the statement: words in the dictionary are matched with the statement, and the slot in the statement where the word matching the dictionary is located is determined as the event slot.

However, using dictionary matching method requires maintaining a data dictionary of event slots and extracting time slots through a strong matching strategy. However, since event slots cannot be exhaustive, they can often only be used to extract some high-frequency event slots. It is difficult to extract low-frequency event slots, thereby reducing the accuracy of event slot extraction.

It can be seen that the event slot extraction method in the related art has a problem of low accuracy in event slot extraction due to the difficulty in exhaustively enumerating event slots.

Contents of the invention

Embodiments of the present disclosure provide an event slot extraction method and device, a storage medium, and an electronic device to at least solve the problem of event slot extraction methods in related technologies due to the difficulty in exhaustively enumerating event slots. The problem of low extraction accuracy.

According to an aspect of an embodiment of the present disclosure, a method for extracting event slots is provided, including: obtaining a target control statement of the event slot to be extracted, wherein the target control statement includes a plurality of words; The control statement performs syntactic analysis to determine a group of candidate words from the plurality of words, wherein the number of candidate words included in the group of candidate words is smaller than the number of words included in the plurality of words; according to the The syntactic relationship between each candidate word in a set of candidate words and the associated words of each candidate word, and the target word is determined from the set of candidate words, wherein the target word and the target word There is a designated syntactic relationship between the related words; the target word and the related words of the target word are determined as the target event slot of the target control statement.

According to another aspect of the embodiment of the present disclosure, an event slot extraction device is also provided, including: an acquisition unit configured to acquire a target control statement of the event slot to be extracted, wherein the target control statement includes a plurality of words; an analysis unit configured to perform syntactic analysis on the target control statement and determine a group of candidate words from the plurality of words, wherein the number of candidate words included in the group of candidate words is less than the number of candidate words. The number of words contained in the plurality of words; the first determination unit is configured to determine the number of words from the set of candidate words based on the syntactic relationship between each candidate word in the set of candidate words and the associated words of each candidate word. The target word is determined among the candidate words, wherein there is a specified syntactic relationship between the target word and the associated words of the target word; the second determination unit is configured to determine the target word and the associated words of the target word. , determined as the target event slot of the target control statement.

According to yet another aspect of the embodiment of the present disclosure, a computer-readable storage medium is also provided. The computer-readable storage medium stores a computer program, wherein the computer program is configured to execute the above event slot during runtime. extraction method.

According to another aspect of the embodiment of the present disclosure, an electronic device is also provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the above-mentioned steps through the computer program. Extraction method of event slot.

In the embodiment of the present disclosure, the candidate words in the control statement are first extracted, and then words belonging to the event slot and their associated words are selected based on the syntactic relationship between the candidate words and associated words. By obtaining the event to be extracted The target control statement of the slot, wherein the target control statement includes multiple words; perform syntactic analysis on the target control statement, and determine a group of candidate words from the multiple words, wherein the candidate words included in the group of candidate words are The number is less than the number of words contained in multiple words; according to the syntactic relationship between each candidate word in the set of candidate words and the associated words of each candidate word, the target word is determined from a set of candidate words, where the target word There is a specified syntactic relationship between the word and the associated words of the target word; the target word and the associated words of the target word are determined as the target event slot of the target control statement. Due to the syntactic analysis of the control statement, the candidates in the control statement are obtained words, according to the syntactic structure formed by the candidate words and the associated words of the candidate words, select words that can be used as event slots, and determine the selected words and their associated words as the extracted event slots, by considering the part of speech, syntax, and sentence structure When extracting event slots using other information features, there is no need to maintain a dictionary corresponding to the event slot. The purpose of accurately extracting event slots can be achieved based on the syntactic structure, achieving the technical effect of improving the accuracy of event slot extraction, and thus solving the related technical problems. The event slot extraction method in has the problem of low accuracy in event slot extraction due to the difficulty in exhaustively enumerating event slots.

Description of drawings

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

In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those of ordinary skill in the art, It is said that other drawings can be obtained based on these drawings without exerting creative labor.

Figure 1 is a schematic diagram of the hardware environment of an optional event slot extraction method according to an embodiment of the present disclosure;

Figure 2 is a schematic flowchart of an optional event slot extraction method according to an embodiment of the present disclosure;

Figure 3 is a schematic diagram of an optional statement structure according to an embodiment of the present disclosure;

Figure 4 is a schematic diagram of another optional statement structure according to an embodiment of the present disclosure;

Figure 5 is a schematic diagram of yet another optional statement structure according to an embodiment of the present disclosure;

Figure 6 is a schematic diagram of yet another optional statement structure according to an embodiment of the present disclosure;

Figure 7 is a schematic diagram of yet another optional statement structure according to an embodiment of the present disclosure;

Figure 8 is a schematic flowchart of another optional event slot extraction method according to an embodiment of the present disclosure;

Figure 9 is a structural block diagram of an optional event slot extraction device according to an embodiment of the present disclosure;

Figure 10 is a structural block diagram of an optional electronic device according to an embodiment of the present disclosure.

Detailed ways

In order to enable those skilled in the art to better understand the present disclosure, the following will clearly and completely describe the technical solutions in the present disclosure embodiments in conjunction with the accompanying drawings. Obviously, the described embodiments are only These are part of the embodiments of this disclosure, not all of them. Based on the embodiments in this disclosure, all other embodiments obtained by those of ordinary skill in the art without creative efforts should fall within the scope of protection of this disclosure.

It should be noted that the terms "first", "second", etc. in the description and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It is to be understood that the data so used are interchangeable under appropriate circumstances so that the embodiments of the disclosure described herein can be practiced in sequences other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, e.g., a process, method, system, product, or apparatus that encompasses a series of steps or units and need not be limited to those explicitly listed. Those steps or elements may instead include other steps or elements not expressly listed or inherent to the process, method, product or apparatus.

According to one aspect of an embodiment of the present disclosure, a method for extracting event slots is provided. The event slot extraction method can be applied to whole-house intelligent digital control application scenarios such as smart home, smart home, smart home device ecology, and intelligent residence (Intelligence House) ecology. Optionally, in this embodiment , the above event slot extraction method can be applied to the hardware environment composed of the terminal 102 and the server 104 as shown in Figure 1 . As shown in Figure 1, the server 104 is connected to the terminal 102 through the network and can be used to provide services (such as application services, etc.) for the terminal or the client installed on the terminal. Cloud computing and/or can be configured on the server or independently of the server. Edge computing services are used to provide data computing services for the server 104.

The above-mentioned network may include but is not limited to at least one of the following: wired network, wireless network. The above-mentioned wired network may include but is not limited to at least one of the following: wide area network, metropolitan area network, and local area network. The above-mentioned wireless network may include at least one of the following: WIFI (Wireless Fidelity, Wireless Fidelity), Bluetooth. The terminal 102 may be, but is not limited to, a PC, a mobile phone, a tablet, a smart air conditioner, a smart hood, a smart refrigerator, a smart oven, a smart stove, a smart washing machine, a smart water heater, a smart washing equipment, a smart dishwasher, a smart projection device, Smart TV, smart clothes drying rack, smart curtains, smart audio and video, smart sockets, smart audio, smart speakers, smart fresh air equipment, smart kitchen and bathroom equipment, smart bathroom equipment, smart sweeping robot, smart window cleaning robot, smart mopping robot, smart Air purification equipment, smart steamers, smart microwave ovens, smart kitchen appliances, smart purifiers, smart water dispensers, smart door locks, etc.

The event slot extraction method in the embodiment of the present disclosure may be executed by the server 104, may be executed by the terminal 102, or may be executed jointly by the server 104 and the terminal 102. Wherein, the terminal 102 may also perform the event slot extraction method of the embodiment of the present disclosure by a client installed thereon.

Taking the server 104 executing the event slot extraction method in this embodiment as an example, Figure 2 is a schematic flowchart of an optional event slot extraction method according to an embodiment of the present disclosure. As shown in Figure 2, The process of this method may include the following steps:

Step S202: Obtain the target control statement of the event slot to be extracted, where the target control statement includes multiple words.

The event slot extraction method in this embodiment can be applied to the scenario of extracting event slots in control statements in the smart home or smart home field, and can be applied to systems with human-computer interaction functions such as voice intelligent question and answer systems. middle. The user inputs a control statement, and the machine extracts the slot of the control statement, thereby analyzing the user's intention and responding to different user intentions. For example, for slot extraction in the schedule field, the slots involved may include slot information such as time, events, singers, and songs.

For different slots, different slot extraction methods can be used. For example, time slots can be extracted using matching rules defined by regular expressions. Singer and song slots can be extracted by maintaining a data dictionary and matching lyrics and music data. Event slots are richer in content. making its extraction more complex.

In order to extract event slots (for example, event slot extraction in the schedule field), the dictionary matching method is usually used to extract event slots from statements. However, due to the difficulty in exhaustively enumerating event slots (for example, event slots in the schedule field) cannot be exhaustive), resulting in low accuracy in event slot extraction. Moreover, even if exhaustive enumeration is possible, if the dictionary library is too large, it will affect the system running speed to a certain extent. Therefore, the strong matching strategy also greatly reduces its generalization.

In addition, algorithm models can also be used to automatically extract event slots, and the model can be trained with large-scale data to automatically extract slots. However, this method is only suitable for slot extraction tasks with a limited number of slots, and has a narrow scope of application. It is more commonly used in tasks with a controllable number of slots, but the generalization controllability of this method is lower than that of the dictionary library. It can be seen that the method of extracting event slots based on dictionary matching and algorithm models cannot take into account the two aspects of generalization and accuracy.

Considering that event slots such as "playing basketball\n" and "going to school" are all verbs plus noun structures, part-of-speech rules can be formulated with the help of part-of-speech analysis to match and extract event slots, and all matches in the sentence can be Phrases of part-of-speech rules are extracted as event slots. For example, the syntactic structure of a verb plus a noun is considered an event slot, and its generalization is stronger than that of the dictionary database.

If only part-of-speech feature information is considered, without considering syntax and sentence structure features, formulating part-of-speech rules to match the way to extract event slots will reduce the accuracy of event slot extraction. For example, because they do not comply with the gerund rules, event slots such as "Go out to play" and "Go to the gym to play basketball" will not be extracted. For another example, "use \v song\n A\n" also satisfies the defined event slot rules, and the song slot is mistakenly extracted as an event slot. This adds a certain degree of difficulty to the semantic understanding of the question and answer system.

In order to improve the accuracy of event slot extraction, in this embodiment, syntactic analysis can be introduced and matching rules can be formulated to assist in extracting event slots. The control statements are parsed by considering syntactic and sentence structure features to determine the corresponding event slot. bit, which can improve the accuracy of event slot extraction.

The target server (an example of the above-mentioned server 104) can obtain the target control statement of the event slot to be extracted, and the target control statement includes multiple words. The above process of obtaining the target control statement of the event slot to be extracted may include receiving the target control statement sent by the terminal device through a communication connection established with the terminal device.

The user can issue a target control statement to his terminal device through voice or other means, and the control statement can be used to control the terminal device to perform the target device operation. After obtaining the target control statement, the terminal device may send the obtained target control statement to the target server. Optionally, what the target server obtains may be a target voice instruction containing a control statement, etc., and extracts the target control statement contained therein by parsing the received target voice instruction.

For example, after the smartphone (the above-mentioned terminal device) obtains the voice command sent by the user "Remind me to go to the gym to play basketball at 8 am tomorrow", it can send the obtained voice command to the server. For receiving voice commands, the server can parse the voice commands and extract the control statements.

It should be noted that the device operation indicated by the above target control instruction may be a device operation performed by the terminal device, or may be a device operation performed by controlling other devices (for example, smart home devices) other than the terminal device. The above-mentioned equipment operation may be an operation of adjusting equipment parameters, or may be an operation of setting a schedule, etc. Correspondingly, the target control instruction may be a schedule reminder instruction. For example, the target control instruction may be an equipment control instruction such as "turn on the water heater tomorrow morning." It can also be a schedule reminder instruction such as "remind me to go to the gym to play basketball at 8 o'clock tomorrow morning", etc., which is not limited in this embodiment.

Step S204: Perform syntax analysis on the target control statement to determine a group of candidate words from multiple words, where the number of candidate words included in a group of candidate words is smaller than the number of words included in the multiple words.

In this embodiment, after acquiring the target control statement, the target server can perform syntax analysis on the target control statement and determine a group of candidate words from multiple words. The number of candidate words included in the group of candidate words is less than a plurality of candidate words. The number of words contained in each word. The extracted group of candidate words can be words with a verb part of speech in the target control sentence, or words with a noun part of speech. When a group of candidate words is a group of candidate verbs, it can be the core verb in the control statement, it can be all the verbs in the control statement, it can be the verb after the core verb in the control statement, or it can be the core verb in the control statement. The previous verb may be a verb whose semantic weight for the control statement exceeds the weight threshold, or may be a verb located at a non-endpoint position of the target control statement in the target control statement, which is not limited in this embodiment.

The above-mentioned process of syntactic analysis of the target control statement may be to use a syntactic analysis algorithm to perform syntactic analysis of the target control statement (which may be implemented using open source syntax analysis software), and extract multiple words included in the target control statement (which may be It is to extract the core words of the sentence), analyze the part-of-speech of each word in multiple words and the relationship between the words (that is, extract the structural features of the sentence), and extract the part-of-speech and sentence structure features of the sentence.

Optionally, after determining the plurality of words included in the target control sentence and the part-of-speech of the words, the target server may determine a group of candidate words based on the part-of-speech of each word in the multiple words, or may first perform a query on the target control sentence. Syntactic analysis determines the core words in the target control sentence, and then determines the words that have a specified syntactic relationship with the core words as candidate words, and then obtains a set of candidate words. It can also use other methods to determine candidate words. This is not limited in the examples.

For example, as shown in Figure 3, when the control statement is "Remind me to go to the gymnasium to play basketball with song A at 8 o'clock tomorrow morning", the "remind", "go" and "play" in the statement can be determined as a set of candidates word.

Step S206: Determine the target word from a group of candidate words based on the syntactic relationship between each candidate word in the group of candidate words and the associated words of each candidate word, where the target word and the associated words of the target word are specifies the syntactic relationship between them.

In this embodiment, after determining a set of candidate words from a plurality of words, the target server may determine words associated with each candidate word in the set of candidate words. The associated words of each of the above candidate words may be adjacent words of each candidate word, or may be words that have a syntactic relationship with each candidate word, or may be other associated relationships with each candidate word. The word, for example, may be the first word after each candidate word, or the first noun after each candidate word, which is not limited in this embodiment.

In this embodiment, after determining the associated words with each candidate word, the target server may determine the target word from a group of candidate words based on the syntactic relationship between each candidate word and the associated words of each candidate word. , optionally, a candidate word whose syntactic relationship between a set of candidate words and associated words is a specified syntactic relationship can be determined as the target word.

For example, specify the syntactic relationship as a verb-object relationship or a parallel relationship, and the associated words of each candidate word are the adjacent words of the candidate word. The candidate words in the control statement "Remind me to go to the gym to play basketball with song A at 8 o'clock tomorrow morning" include : "Reminder", "Go", "Hit", "Reminder" adjacent words are "I", the syntactic relationship formed between the two is neither a verb-object relationship nor a parallel relationship, therefore, "Remind" is not the target word , the word adjacent to "go" is "gymnasium", and the syntactic relationship formed between the two is a verb-object relationship. "Go" is the target word, and the word adjacent to "play" is "basketball", and the relationship formed between the two is The syntactic relationship of is also a verb-object relationship, and "beat" is also a target word.

Step S208: Determine the target word and related words of the target word as the target event slot of the target control statement.

In this embodiment, after determining the target word, the target server may determine the target word and words associated with the target word as the target event slot of the target control statement. Optionally, when there are multiple target words, each target word and associated words of each target word may be determined as the target event slot of the target control statement.

For example, when the server determines that the target words include "go" and "play" from the control statement "Remind me to go to the gymnasium to play basketball with song A at 8 o'clock tomorrow morning", it can add both "go to the gym" and "play basketball". Determine the event slot of this control statement.

Through the above steps, the target control statement of the event slot to be extracted is obtained, where the target control statement includes multiple words; the target control statement is syntactically analyzed to determine a group of candidate words from the multiple words, wherein a group of The number of candidate words contained in a candidate word is smaller than the number of words contained in multiple words; based on the syntactic relationship between each candidate word in a set of candidate words and the associated words of each candidate word, from a set of candidate words The target word is determined, in which the target word and the associated words of the target word are designated syntactic relationships; the target word and the associated words of the target word are determined as the target event slot of the target control statement, solving the problems in related technologies The event slot extraction method has the problem of low accuracy in event slot extraction due to the difficulty in exhaustively enumerating event slots. This method improves the accuracy of event slot extraction.

In an exemplary embodiment, the target word is determined from a set of candidate words based on the syntactic relationship between each candidate word in the set of candidate words and associated words of each candidate word, including:

S11, determine the target verb from a group of candidate verbs based on the syntactic relationship between each candidate verb in the group of candidate verbs and the associated words of each candidate verb, where the group of candidate words includes a group of candidate verbs, Target words include target verbs.

Since event slots in control statements generally contain verbs, after performing syntactic analysis on the target control statement and determining a group of candidate words from multiple words, the target verb can be determined first among a group of candidate words, and then According to the target verb, determine the target event slot of the target control statement. Optionally, the above-mentioned process of determining the target verb in a group of candidate words may be: based on the syntactic relationship between each candidate verb in the group of candidate verbs and the associated words of each candidate verb, from a group of candidate verbs The target verb is determined, the above group of candidate words includes a group of candidate verbs, and the target word includes the target verb.

The above-mentioned syntactic relations may be specified syntactic relations or other syntactic relations. Optionally, a candidate verb in a set of candidate verbs whose syntactic relationship with the associated word is a specified syntactic relationship may be determined as the target verb.

For example, when the syntactic relationship is specified as a verb-object relationship or a parallel relationship, and the associated words of each candidate verb are adjacent words of the candidate verb, the candidates in the control statement "Remind me to go to the gym to play basketball with song A at 8 o'clock tomorrow morning" Verbs include: "remind", "go", "beat", the adjacent word of "remind" is "I", the syntactic relationship formed between the two is neither a verb-object relationship nor a parallel relationship, therefore, "remind" is not The target verb, the adjacent word of "go" is "gymnasium", and the syntactic relationship formed between the two is a verb-object relationship. "Go" is the target verb, and the adjacent word of "play" is "basketball". The syntactic relationship formed between them is also a verb-object relationship, and "beat" is also the target verb.

Through this embodiment, the target verb is determined from the control statement according to the syntactic relationship between each candidate verb in the control statement and the associated words of each candidate verb, which can improve the efficiency of determining the target verb.

In an exemplary embodiment, syntactic analysis is performed on the target control statement to determine a set of candidate words from multiple words, including:

S21, perform syntactic analysis on the target control statement, and determine multiple words contained in the target control statement and the part-of-speech of each word in the multiple words;

S22: Determine each word whose part-of-speech is a verb among the plurality of words as a candidate verb, and obtain a set of candidate verbs.

In this embodiment, the target server may perform syntax analysis on the target control statement, and determine multiple words included in the target control sentence and the part-of-speech of each word in the multiple words. Optionally, in addition to the part of speech of each word, the target server may also determine a syntactic relationship between each word and words other than the word among the plurality of words. For example, the server can perform syntactic analysis on the control statement "Remind me to go to the gymnasium to play basketball with song A at 8 o'clock tomorrow morning" and obtain the part of speech of each word in the multiple words included in the control statement, as well as the relationship between each word and other words. syntactic relationship between them.

The above-mentioned process of performing syntactic analysis on the target control sentence and determining the multiple words contained in the target control sentence and the part-of-speech of each word in the multiple words may be: analyzing the target control sentence through a part-of-speech tagging algorithm to obtain that the target control sentence contains of multiple words and the part of speech for each of the multiple words. The above-mentioned part-of-speech tagging algorithm may include at least one of the following: sequence model, hidden Markov model (Hidden Markov Model, HMM), Markov model, for example, MEMM (Maximum Entropy Markov Model, maximum entropy Markov model), CRFS (Conditional Random Fields, Conditional Random Fields) and other Markov models in a broad sense, deep learning, for example, conventional classifiers of machine learning (for example, SVM (Support Vector Machine, Support Vector Machine), to RNN (Recurrent Neural Network, Deep learning algorithms represented by recurrent neural networks), etc. are not limited in this embodiment.

For example, as shown in Figures 3, 4, 5, 6 and 7, different control statements can be syntactically analyzed to determine the words contained in the different control statements, the parts of speech of the words, and the syntactic relationships between the words. , among them, nt (temporal noun, time noun), p (preposition, time noun), n (general noun, general noun), v (verb, verb), r (pronoun, pronoun), wp (punctuation, Punctuation), nl (location noun, location noun), nd (direction noun, direction noun) are used to indicate the part of speech of the corresponding word, and control the HED (head, core word), ATT (attribute, centering relationship), ADV above the statement (adverbial, adverbial structure), POB (preposition-object, preposition-object relationship), DBL (double, concurrent language), VOB (verb-object, verb-object relationship), COO (coordinate, parallel relationship), SBV (subject- verb (subject-predicate relationship) and WP (punctuation, punctuation) are used to express the syntactic relationship between words, and arrows are used to express the directional relationship between words.

After determining the multiple words contained in the target control statement and the part-of-speech of each word, the target server can determine each word whose part-of-speech is a verb among the multiple words as a candidate verb, and obtain a set of candidate verbs, that is, a set of The candidate verbs are all verbs in multiple words.

Through this embodiment, words whose part-of-speech is verb in the control statement are determined as candidate verbs, which can simplify the determination steps of candidate verbs and improve the efficiency of determining candidate verbs.

S31, perform syntactic analysis on the target control statement and determine the core verb of the target control instruction;

S32, determine each verb in multiple words that has a specified syntactic relationship with the core verb as a candidate verb, and obtain a set of candidate verbs;

S33, find verbs that have a specified syntactic relationship with any candidate verb in a set of candidate verbs;

S34: When a verb having a specified syntactic relationship with any candidate verb is found, add the found verb as a candidate verb to a group of candidate verbs.

In this embodiment, a set of candidate verbs may be determined based on core verbs in the target control statement. The target server can perform syntax analysis on the target control statement and determine the core verb of the target control instruction. The above-mentioned core verbs are verbs that dominate other words in the target control instructions and are not dominated by other words. The core verb represents the core HED of the entire sentence (root). For example, the core verbs in the schedule field are generally "remind", "call", etc. The position of the core word in the sentence has certain reference value for event slot extraction; based on the core The syntactic structure composed of verbs and adjacent words of the core verb is used to extract event slots. Since syntactic and sentence structure features are taken into account, the accuracy of event slot extraction can be improved.

For example, the words pointed to by root in Figures 3, 4, 5, 6 and 7 are the core verbs corresponding to the control statements. In the control statement shown in Figure 3, "remind" is the part of the control statement. Core words. In the control statement shown in Figure 4, "go" is the core word in the control statement. In the control statement shown in Figure 5, "remind" is the core word in the control statement, as shown in Figure 6 In the control statement shown, "remind" is the core word in the control statement. In the control statement shown in Figure 7, "remember" is the core word in the control statement.

After determining the core verb of the target control instruction, the target server can determine each verb among the multiple words included in the target control statement that has a specified syntactic relationship with the core verb as a candidate verb, and obtain a set of candidate verbs. , the above specified syntactic relationship may be a verb that has a VOB relationship with the core verb, or it may be a verb that has a COO relationship with the core verb, which is not limited in this embodiment.

The above group of candidate verbs may include multiple candidate verbs, or may include only one candidate verb, which is not limited in this embodiment. For example, as shown in Figure 3, after determining that "remind" is the core word of the control statement, the verb "go" that has a VOB relationship with "remind" can be determined as a candidate verb, because in the control statement, in addition to the candidate After the verb "go", there is no verb with a VOB relationship or a COO relationship with the core verb. Therefore, the set of candidate verbs in the control statement only includes the candidate verb "go".

After determining a set of candidate verbs, the target server may look for verbs that have a specified syntactic relationship with any candidate verb in the set of candidate verbs. Optionally, the above-mentioned process of finding verbs that have a specified syntactic relationship with any candidate verb in a group of candidate verbs may be: finding verbs that have a VOB relationship with any candidate verb in a group of candidate verbs or verbs that have a COO relationship. verb. This is not limited in this embodiment. When a verb with a specified syntactic relationship to any candidate verb is found, the found verb can be added as a candidate verb to a set of candidate verbs.

For example, as shown in Figure 3, after determining the candidate verb "go", since there is a COO relationship between "beat" and the candidate verb "go", "beat" can also be added to a group of candidate verbs. At this time The number of verbs included in a group of candidate verbs has changed from 1 (i.e. "to") to 2 (i.e. "beat", "go"). As shown in Figure 4, after determining the candidate verb "go" and determining the verbs "hit" and "remember" that have a COO relationship with "remind" as a group of candidate verbs, since the verb "remind" is different from the candidate verb The verb "remember" has a VOB relationship, so "remind" can also be added to a group of candidate verbs. At this time, the number of verbs included in a group of candidate verbs changes from 2 (i.e., "beat", "remember") to 3 (i.e. "beat", "remember", "remind").

Through this embodiment, the verbs that have a specified syntactic relationship with the core verb in the control statement are determined as candidate verbs, and the verbs that have a specified syntactic relationship with the candidate verbs are also determined as candidate verbs, which can improve the determination of candidate verbs. accuracy to avoid missing candidate verbs.

In an exemplary embodiment, the associated words of each candidate verb are adjacent words of each candidate verb, and the adjacent words of each candidate verb are located after each candidate verb and adjacent to each candidate verb. Words; determine the target verb from a set of candidate verbs based on the syntactic relationship between each candidate verb in the set of candidate verbs and the associated words of each candidate verb, including:

S41: Determine the candidate verb in a group of candidate verbs that satisfies the first filtering condition between adjacent words as the target verb, where the first filtering condition includes a syntactic relationship, which is a specified syntactic relationship.

In this embodiment, after determining a group of candidate verbs, the target server may determine a candidate verb among the group of candidate verbs that satisfies the first filtering condition between adjacent words as the target verb.

Optionally, the associated words of each candidate verb are adjacent words of each candidate verb, and the adjacent words of each candidate verb are words located after each candidate verb and adjacent to each candidate verb. A filter condition including syntactic relations is the specified syntactic relation.

For example, as shown in Figure 4, in the control statement "I am going to play basketball tomorrow, remember to remind me at eight o'clock tomorrow morning", the candidate verbs "play" and "remember", and the associated words of "remind" are "basketball" , "at" and "I" instead of "to", "," and "eight o'clock".

Through this embodiment, the candidate words that have a specified syntactic relationship with adjacent words in a group of candidate verbs are determined as target words, which can improve the accuracy and flexibility of determining the target words.

In an exemplary embodiment, a candidate verb among a group of candidate verbs that satisfies the first filtering condition between adjacent words is determined as the target verb, including:

S51, when the syntactic relationship between any candidate verb in a group of candidate verbs and the adjacent words of any candidate verb is a verb-object relationship, and the part of speech of any adjacent word of the candidate verb is a noun. , determine any candidate verb as the target verb, where the specified syntactic relationship includes a verb-object relationship, and the first filtering condition also includes the part-of-speech of the adjacent word being a noun;

S52: When the syntactic relationship between any candidate verb and the adjacent words of any candidate verb is a parallel relationship, determine any candidate verb as the target verb, where the specified syntactic relationship includes a parallel relationship.

In this embodiment, the above-mentioned first filtering condition may include but is not limited to one of the following: the syntactic relationship between the candidate verb and the adjacent words is a verb-object relationship, and the part of speech between the candidate verb and the adjacent words is a noun; The syntactic relationship between the candidate verb and adjacent words is a parallel relationship.

As an optional implementation, the syntactic relationship between any candidate verb in a group of candidate verbs and the adjacent words of any candidate verb can be a verb-object relationship, and the adjacent words of any candidate verb When the part-of-speech of the word is a noun, any candidate verb is determined as the target verb. The above-mentioned specified syntactic relationship includes a verb-object relationship (ie, VOB relationship). The above-mentioned first filtering condition also includes that the part-of-speech of the adjacent word is a noun.

For example, as shown in Figure 4, in the control statement "I am going to play basketball tomorrow, remember to remind me at eight o'clock tomorrow morning", the adjacent words of the candidate verbs "play", "remember" and "remind" are respectively " "Basketball", "in" and "I". Among them, the syntactic relationship formed between the candidate verb "play" and "basketball" is a verb-object relationship, and "basketball" is a noun, then the candidate verb "play" can be determined as There is no syntactic relationship between the target verb and the candidate verbs "remember" and "in", so the candidate verb "remember" is not the target verb. Although the syntactic relationship between the candidate verb "remind" and "I" is a verb-object relationship , but the part of speech of "I" is a pronoun, not a noun, so the candidate verb "remind" is not the target verb.

As another optional implementation, any candidate verb can be determined as the target verb when the syntactic relationship between any candidate verb and the adjacent words of any candidate verb is a parallel relationship, where the syntax is specified Relationships include parallel relationships (ie, COO relationships).

For example, as shown in Figure 7, in the control statement "Remember to remind me to go for a walk at eight o'clock tomorrow morning", the adjacent words of the candidate verbs "remind", "go out" and "walk" are "I" and "walk" respectively. ", among which, the syntactic relationship formed between the candidate verb "remind" and "I" is that DBL is not a verb-object relationship or a parallel relationship, so the candidate verb "remind" is not the target verb, and the candidate verb "go out" and "walk" The syntactic relationship formed between them is a parallel relationship, so the candidate verb "go out" can be determined as the target verb.

Through this embodiment, the target verb is determined from a group of candidate verbs according to the syntactic relationship between any candidate verb in the group of candidate verbs and the adjacent words of any candidate verb and the part of speech of the adjacent words of any candidate verb. , which can improve the accuracy of determining the target verb, thereby improving the accuracy of the determined event slot.

In an exemplary embodiment, the associated words of each candidate verb are words that have a syntactic relationship with each candidate verb; according to the syntax between each candidate verb in a set of candidate verbs and the associated words of each candidate verb Relationship, determine the target verb from a set of candidate verbs, including:

S61: Determine a candidate verb in a group of candidate verbs that satisfies a second filtering condition with a word that has a syntactic relationship as a target verb, where the second filtering condition includes a syntactic relationship that is a specified syntactic relationship.

In this embodiment, after determining a group of candidate verbs, the target server may determine a candidate verb among the group of candidate verbs that satisfies the second filtering condition with words that have a syntactic relationship as the target verb.

Optionally, the associated words of each of the above candidate verbs are words that have a syntactic relationship with each candidate verb, and the second filtering condition includes that the syntactic relationship is a specified syntactic relationship.

For example, a candidate verb whose syntactic relationship between a group of candidate verbs and a word that has a syntactic relationship is a VOB relationship or has a COO relationship can be determined as the target verb.

Through this embodiment, a candidate word with a specified syntactic relationship between a group of candidate verbs and a word with a syntactic relationship is determined as the target word, which can improve the accuracy and flexibility of determining the target word. .

In an exemplary embodiment, a candidate verb that satisfies the second filtering condition among a group of candidate verbs and words that have a syntactic relationship is determined as the target verb:

S71, the syntactic relationship between any candidate verb in a group of candidate verbs and the word that has a syntactic relationship with any candidate verb is a verb-object relationship, and the part of speech of the word that has a syntactic relationship with any candidate verb is a noun In the case of , any candidate verb is determined as the target verb, wherein the specified syntactic relationship includes a verb-object relationship, and the second filtering condition also includes that the part-of-speech of the word with the syntactic relationship is a noun;

S72, in the case where the syntactic relationship between any candidate verb and a word having a syntactic relationship with any candidate verb is a parallel relationship, determine any candidate verb as the target verb, where the specified syntactic relationship includes a parallel relationship.

In this embodiment, the above-mentioned first filtering condition may include but is not limited to one of the following: the syntactic relationship between the candidate verb and the words that have a syntactic relationship with the candidate verb is a verb-object relationship, and the candidate verb has an adjacent word The part of speech of is a noun; the syntactic relationship between the candidate verb and the words that have a syntactic relationship with the candidate verb is a parallel relationship.

As an optional implementation, the syntactic relationship between any candidate verb in a set of candidate verbs and the word that has a syntactic relationship with any candidate verb is a verb-object relationship, and has a syntactic relationship with any candidate verb. When the part-of-speech of the related word is a noun, any candidate verb is determined as the target verb. The above-mentioned specified syntactic relationship includes a verb-object relationship (i.e., VOB relationship). The above-mentioned second filtering condition also includes the part-of-speech of the word with the syntactic relationship. as a noun.

For example, as shown in Figure 4, in the control statement "I am going to play basketball tomorrow, remember to remind me at eight o'clock tomorrow morning", the candidate verbs "go", "play", "remember" and "remind" are associated words They are "go", "basketball", "remind" and "I". Among them, the candidate verb "go" has no word with verb-object relationship or parallel relationship except "beat", and "beat" is also related to "beat". There is a verb-object relationship between "basketball", so "go" is not the target verb, there is a verb-object relationship between the candidate verb "play" and "basketball", and "basketball" is a noun, so the candidate verb "play" can be determined As the target verb, the candidate verb "remember" has a parallel relationship with "remind". However, since "remind" not only has the candidate verb "remember", it also has a verb-object relationship with "I", so the candidate verb "remember" It is not the target verb. The candidate verb "remind" has a verb-object relationship with "I". However, the part of speech of "I" is a pronoun, not a noun, so the candidate verb "remind" is not the target verb.

As another optional implementation, when the syntactic relationship between any candidate verb and a word that has a syntactic relationship with any candidate verb is a parallel relationship, any candidate verb can be determined as the target verb, as specified above Syntactic relations include parallel relations (i.e., COO relations).

For example, as shown in Figure 7, in the control statement "Remember to remind me to go for a walk at eight o'clock tomorrow morning", the associated words of the candidate verbs "remind", "go out" and "walk" are "I" and "walk" respectively. , among them, the syntactic relationship formed between the candidate verb "remind" and "I" is that DBL is not a verb-object relationship or a parallel relationship, so the candidate verb "remind" is not the target verb, and the candidate verb "go out" and "walk" The syntactic relationship formed between them is a parallel relationship, so the candidate verb "go out" can be determined as the target verb.

It should be noted that the gerund structure of "go to the gym\n" or "play basketball\n" is called the verb-object relationship VOB in syntax, and the verb-object relationship VOB structure between two adjacent positions is in Syntactically it is called the parallel relationship COO. With the help of this parallel structure in the object-verb relationship, the correct event slot can be extracted: "Go to the gym to play basketball", and the event slot extraction accuracy is higher.

Through this embodiment, a group of candidate verbs is determined from a group of candidate verbs based on the syntactic relationship between any candidate verb in the group of candidate verbs and the words that have a syntactic relationship with any candidate verb, and the part of speech of the associated words of any candidate verb. The target verb can improve the accuracy of determining the target verb, thereby improving the accuracy of determining the event slot.

In an exemplary embodiment, the target word and the associated words of the target word are determined as the target event slot of the target control statement, including:

S81, when there are multiple target words, determine each target word and the associated words of each target word as event slots corresponding to each target word;

S82: Determine the combination of event slots corresponding to each target word as the target event slot.

In this embodiment, when there are multiple target words, the target server can determine the event slot corresponding to each target word and the associated words of each target word, and add the event slot corresponding to each target word. The combination of event slots corresponding to target words is determined as the target event slot.

For example, when the control statement is "Remind me to go to the gymnasium to play basketball with song A at 8 o'clock tomorrow morning", the target words "go", "play" and the words "gymnasium" and "basketball" associated with the target words can be determined as Event slot, the final event slot obtained is: "Go to the gym to play basketball", rather than a single "Play basketball", so that the slot extraction is more complete and comprehensive, making the question and answer system in NLG (Natural Language Generation, natural language generation) ) broadcast reply, the user experience is better.

The above-mentioned process of determining the combination of event slots corresponding to each target word as the target event slot may be to combine the adjacent event slots when the event slots corresponding to each target word are adjacent. Splice them into one event slot, and then determine the spliced event slot as the target event slot.

Optionally, when the target control statement is a schedule setting statement corresponding to the target schedule, the target server can determine the schedule reminder time corresponding to the target schedule from the target control statement. For example, when the control statement is "tomorrow at 8 a.m. When "remind me to go to the gym to play basketball with song A", you can determine 8 a.m. tomorrow as the schedule reminder time corresponding to the schedule.

After determining the target event slot, the target server can construct a schedule reminder statement according to the target event slot. The target reminder statement is used to remind the execution of the target schedule. The above-mentioned process of constructing a schedule reminder sentence according to the target event slot may be: according to the target event slot, extract the words located in the target event slot from the target control sentence, and construct a schedule reminder sentence based on these words. In this embodiment, this No restrictions. , after constructing the schedule reminder sentence, when the target reminder time arrives, the target reminder sentence can be converted into the corresponding reminder voice for voice broadcast. Here, the target reminder time is the reminder time of the target schedule extracted from the target control sentence. .

Through this embodiment, the combination of event slots corresponding to each target word is determined as the event slot, which can improve the accuracy of determining the event slot.

The method for extracting event slots in the embodiment of the present disclosure will be explained below in conjunction with optional examples. This optional example provides an optimization solution that uses syntactic analysis to extract slots for schedule items in the schedule field. In the process of extracting event slots in the schedule field of the question and answer system, syntactic analysis is introduced to assist in extracting event slots in the schedule field. By extracting the core verb of the sentence, using the core verb as a reference, we use the verb-object relationship of the verb immediately adjacent to the noun and the parallel relationship of the verb immediately adjacent to the verb to formulate a syntactic rule strategy for event slot extraction to improve the accuracy of event slot extraction and enhance Generalizability of schedule item slotting.

Combined with Figure 8, the process of the event slot extraction method in this optional example may include the following steps:

Step S802: parse the sentence by calling a syntax analysis algorithm to obtain candidate verbs in the sentence.

Call the syntactic analysis algorithm to analyze the syntactic structure of the sentence, and you can get the candidate verbs in the sentence structure, and other words can diverge around the core verb.

Step S804: Determine the part-of-speech of adjacent words of the candidate verb and the structural relationship formed between the adjacent words and the candidate verb.

Step S806: When the syntactic relationship formed between the adjacent words and the candidate verb is a verb-object relationship or a parallel relationship, the slot where the candidate verb and the words adjacent to the candidate verb are located are determined as event slots.

When the syntactic structure composed of a verb and a noun immediately adjacent is a verb-object structure VOB (corresponding to the above verb-object relationship), if the two VOB structures are also adjacent, the final event slot is the assembly of the two VOBs, for example, "Go to the gym to play basketball." , if there is no verb-object structure in the sentence, but there is a COO parallel relationship consisting of a verb immediately adjacent to the verb, it is also considered to be a possibility of the matter slot, such as "go out for a walk".

Through this optional example, syntactic analysis technology is applied to the event slot extraction of complex sentences. The position of the core verb in the sentence, the verb-object structure VOB in the sentence, and the parallel relationship COO can all provide information for the extraction of event slots in the schedule field. Certain auxiliary information makes the slot extraction algorithm more robust and accurate. The introduction of syntactic structure feature information enables more accurate extraction of event slots in complex slots such as schedule matters.

It should be noted that for the sake of simple description, the foregoing method embodiments are expressed as a series of action combinations. However, those skilled in the art should know that the present disclosure is not limited by the described action sequence. Because in accordance with the present disclosure, certain steps may be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily necessary for the present disclosure.

Through the description of the above embodiments, those skilled in the art can clearly understand that the method according to the above embodiments can be implemented by means of software plus the necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is Better implementation. Based on this understanding, the technical solution of the present disclosure can be embodied in the form of a software product in essence or that contributes to the existing technology. The computer software product is stored in a storage medium (such as ROM (Read-Only Memory, Read-only memory)/RAM (Random Access Memory, disk, optical disk), including a number of instructions to make a terminal device (can be a mobile phone, computer, server, or network device, etc.) to execute this Methods described in various embodiments are disclosed.

According to another aspect of the embodiment of the present disclosure, an event slot extraction device for implementing the above event slot extraction method is also provided. Figure 9 is a structural block diagram of an optional event slot extraction device according to an embodiment of the present disclosure. As shown in Figure 9, the device may include:

The acquisition unit 902 is configured to acquire the target control statement of the event slot to be extracted, where the target control statement includes multiple words;

The analysis unit 904 is connected to the acquisition unit 902 and is configured to perform syntactic analysis on the target control statement and determine a group of candidate words from a plurality of words, wherein the number of candidate words contained in a group of candidate words is less than the number of candidate words contained in the plurality of words. the number of words;

The first determination unit 906 is connected to the analysis unit 904 and is configured to determine the target word from a group of candidate words based on the syntactic relationship between each candidate word in the group of candidate words and the associated words of each candidate word, Among them, there is a specified syntactic relationship between the target word and the associated words of the target word;

The second determination unit 908 is connected to the first determination unit 906 and is configured to determine the target word and related words of the target word as the target event slot of the target control statement.

It should be noted that the acquisition unit 902 in this embodiment can be configured to perform the above step S202, the analysis unit 904 in this embodiment can be configured to perform the above step S204, and the first determination unit 906 in this embodiment can be configured to The above-mentioned step S206 is executed, and the second determination unit 908 in this embodiment may be configured to execute the above-mentioned step S208.

Through the above module, the target control sentence of the event slot to be extracted is obtained, where the target control sentence includes multiple words; the target control sentence is syntactically analyzed to determine a set of candidate words from the multiple words, among which a set of The number of candidate words contained in a candidate word is smaller than the number of words contained in multiple words; based on the syntactic relationship between each candidate word in a set of candidate words and the associated words of each candidate word, from a set of candidate words The target word is determined, in which the target word and the associated words of the target word are designated syntactic relationships; the target word and the associated words of the target word are determined as the target event slot of the target control statement, solving the problems in related technologies The event slot extraction method has the problem of low accuracy in event slot extraction due to the difficulty in exhaustively enumerating event slots. This method improves the accuracy of event slot extraction.

In an exemplary embodiment, the first determining unit includes:

The first determination module is configured to determine the target verb from a group of candidate verbs based on the syntactic relationship between each candidate verb in the group of candidate verbs and associated words of each candidate verb, where the group of candidate words includes A set of candidate verbs, the target word includes the target verb.

In an exemplary embodiment, the analysis unit includes:

The second determination module is configured to perform syntactic analysis on the target control statement and determine multiple words contained in the target control statement and the part-of-speech of each word in the multiple words;

The third determination module is configured to determine each word whose part-of-speech is a verb among the plurality of words as a candidate verb, and obtain a group of candidate verbs.

In an exemplary embodiment, the analysis unit includes:

The analysis module is configured to perform syntactic analysis on the target control statement and determine the core verb of the target control instruction;

The fourth determination module is configured to determine each verb in multiple words that has a specified syntactic relationship with the core verb as a candidate verb, and obtain a set of candidate verbs;

A search module configured to search for verbs that have a specified syntactic relationship with any candidate verb in a set of candidate verbs;

The add module is configured to add the found verb as a candidate verb to a set of candidate verbs when a verb having a specified syntactic relationship with any candidate verb is found.

In an exemplary embodiment, the associated words of each candidate verb are adjacent words of each candidate verb, and the adjacent words of each candidate verb are located after each candidate verb and adjacent to each candidate verb. Word; the first determination module includes:

The first determination sub-module is configured to determine candidate verbs from a group of candidate verbs that meet the first filtering condition between adjacent words as the target verb, where the first filtering condition includes a syntactic relationship, which is a specified syntactic relationship.

In an exemplary embodiment, the first determination sub-module includes:

The first determination subunit is set so that the syntactic relationship between any candidate verb in a group of candidate verbs and the adjacent words of any candidate verb is a verb-object relationship, and the syntactic relationship between the adjacent words of any candidate verb is When the part-of-speech is a noun, any candidate verb is determined as the target verb, where the specified syntactic relationship includes a verb-object relationship, and the first filtering condition also includes that the part-of-speech of the adjacent word is a noun;

The second determination subunit is configured to determine any candidate verb as the target verb when the syntactic relationship between any candidate verb and the adjacent words of any candidate verb is a parallel relationship, wherein the specified syntactic relationship includes parallelism relation.

In an exemplary embodiment, the associated words of each candidate verb are words that have a syntactic relationship with each candidate verb; the first determination module includes:

The second determination submodule is configured to determine candidate verbs from a group of candidate verbs that meet the second filtering condition with words that have syntactic relationships as target verbs, where the second filtering condition includes a syntactic relationship that is a specified syntactic relationship.

In an exemplary embodiment, the second determination sub-module includes:

The third determination subunit is set so that the syntactic relationship between any candidate verb in a group of candidate verbs and the word that has a syntactic relationship with any candidate verb is a verb-object relationship, and has a syntactic relationship with any candidate verb. When the part-of-speech of the word is a noun, any candidate verb is determined as the target verb, where the specified syntactic relationship includes a verb-object relationship, and the second filtering condition also includes the part-of-speech of the word with the syntactic relationship being a noun;

The fourth determination subunit is configured to determine any candidate verb as the target verb when the syntactic relationship between any candidate verb and a word that has a syntactic relationship with any candidate verb is a parallel relationship, wherein the syntactic relationship is specified Including parallel relationships.

In an exemplary embodiment, the second determining unit includes:

The fifth determination module is configured to determine each target word and associated words of each target word as event slots corresponding to each target word when there are multiple target words;

The sixth determination module is configured to determine the combination of event slots corresponding to each target word as the target event slot.

It should be noted here that the examples and application scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the contents disclosed in the above embodiments. It should be noted that the above module, as part of the device, can run in the hardware environment as shown in Figure 1, and can be implemented by software or hardware, where the hardware environment includes a network environment.

According to yet another aspect of the embodiments of the present disclosure, a storage medium is also provided. Optionally, in this embodiment, the above storage medium can be used to execute the program code of any of the above event slot extraction methods in the embodiment of the present disclosure.

Optionally, in this embodiment, the above storage medium may be located on at least one network device among multiple network devices in the network shown in the above embodiment.

Optionally, in this embodiment, the storage medium is configured to store program codes for performing the following steps:

S1, obtain the target control statement of the event slot to be extracted, where the target control statement includes multiple words;

S2, perform syntactic analysis on the target control statement, and determine a set of candidate words from multiple words, where the number of candidate words included in a set of candidate words is smaller than the number of words included in multiple words;

S3: Determine the target word from a set of candidate words based on the syntactic relationship between each candidate word in the set of candidate words and the associated words of each candidate word, where the relationship between the target word and the associated words of the target word To specify syntactic relations;

S4: Determine the target word and related words of the target word as the target event slot of the target control statement.

Optionally, for specific examples in this embodiment, reference may be made to the examples described in the above embodiments, which will not be described again in this embodiment.

Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: U disk, ROM, RAM, mobile hard disk, magnetic disk or optical disk and other various media that can store program codes.

According to yet another aspect of the embodiments of the present disclosure, an electronic device for implementing the above event slot extraction method is also provided. The electronic device may be a server, a terminal, or a combination thereof.

Figure 10 is a structural block diagram of an optional electronic device according to an embodiment of the present disclosure. As shown in Figure 10, it includes a processor 1002, a communication interface 1004, a memory 1006 and a communication bus 1008. The processor 1002, the communication interface 1004 and memory 1006 complete communication with each other through communication bus 1008, where,

Memory 1006 configured to store computer programs;

When the processor 1002 is configured to execute the computer program stored on the memory 1006, it implements the following steps:

Optionally, in this embodiment, the communication bus may be a PCI (Peripheral Component Interconnect, Peripheral Component Interconnect Standard) bus, or an EISA (Extended Industry Standard Architecture, Extended Industry Standard Architecture) bus, etc. The communication bus can be divided into address bus, data bus, control bus, etc. For ease of presentation, only one thick line is used in Figure 10, but it does not mean that there is only one bus or one type of bus. The communication interface is used for communication between the above-mentioned electronic device and other equipment.

The memory may include RAM or non-volatile memory, such as at least one disk memory. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

As an example, the memory 1006 may include, but is not limited to, the acquisition unit 902, the analysis unit 904, the first determination unit 906 and the second determination unit 908 in the event slot extraction device. In addition, it may also include but is not limited to other module units in the extraction device of the above-mentioned event slots, which will not be described again in this example.

The above-mentioned processor can be a general-purpose processor, which can include but is not limited to: CPU (Central Processing Unit, central processing unit), NP (Network Processor, network processor), etc.; it can also be a DSP (Digital Signal Processing, digital signal processor) ), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array, field programmable gate array) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

Those of ordinary skill in the art can understand that the structure shown in Figure 10 is only illustrative, and the device that implements the above event slot extraction method can be a terminal device, and the terminal device can be a smart phone (such as an Android phone, iOS phone, etc.), Tablet computers, handheld computers, and mobile Internet devices (Mobile Internet Devices, MID), PAD and other terminal devices. FIG. 10 does not limit the structure of the above-mentioned electronic device. For example, the electronic device may also include more or fewer components (such as network interfaces, display devices, etc.) than shown in FIG. 10 , or have a different configuration than that shown in FIG. 10 .

Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program. The program can be stored in a computer-readable storage medium, and the storage medium can Including: flash disk, ROM, RAM, magnetic disk or optical disk, etc.

The above serial numbers of the embodiments of the present disclosure are only for description and do not represent the advantages and disadvantages of the embodiments.

If the integrated units in the above embodiments are implemented in the form of software functional units and sold or used as independent products, they can be stored in the above computer-readable storage medium. Based on this understanding, the technical solution of the present disclosure is essentially or contributes to the existing technology, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, It includes several instructions to cause one or more computer devices (which can be personal computers, servers or network devices, etc.) to execute all or part of the steps of the methods described in various embodiments of the present disclosure.

In the above-mentioned embodiments of the present disclosure, each embodiment is described with its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

In the several embodiments provided by this disclosure, it should be understood that the disclosed client can be implemented in other ways. Among them, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or may be Integrated into another system, or some features can be ignored, or not implemented. On the other hand, the coupling or direct coupling or communication connection between each other shown or discussed may be through some interfaces, and the indirect coupling or communication connection of the units or modules may be in electrical or other forms.

The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution provided in this embodiment.

In addition, each functional unit in various embodiments of the present disclosure may be integrated into one processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit. The above integrated units can be implemented in the form of hardware or software functional units.

The above are only preferred embodiments of the present disclosure. It should be pointed out that for those of ordinary skill in the art, several improvements and modifications can be made without departing from the principles of the present disclosure. These improvements and modifications can also be made. should be regarded as the scope of protection of this disclosure.

Claims

A method for extracting event slots, including:

Obtain the target control statement of the event slot to be extracted, wherein the target control statement includes multiple words;

Perform syntactic analysis on the target control statement to determine a group of candidate words from the plurality of words, wherein the number of candidate words included in the group of candidate words is less than the number of words included in the plurality of words. ;

A target word is determined from the set of candidate words according to the syntactic relationship between each candidate word in the set of candidate words and an associated word of each candidate word, wherein the target word is related to the There is a specified syntactic relationship between related words describing the target word;

The target word and related words of the target word are determined as the target event slot of the target control statement.
The method according to claim 1, wherein the syntactic relationship between each candidate word in the set of candidate words and associated words of each candidate word is selected from the set of candidate words. Identify target words, including:

A target verb is determined from the set of candidate verbs based on the syntactic relationship between each candidate verb in the set of candidate verbs and an associated word of each candidate verb, wherein the set of candidate words includes all Describe a group of candidate verbs, and the target word includes the target verb.
The method according to claim 2, wherein the step of performing syntactic analysis on the target control statement and determining a group of candidate words from the plurality of words includes:

Perform syntactic analysis on the target control statement to determine the plurality of words contained in the target control statement and the part-of-speech of each word in the plurality of words;

Each word whose part-of-speech is a verb among the plurality of words is determined as a candidate verb, and the group of candidate verbs is obtained.
The method according to claim 2, wherein the step of performing syntactic analysis on the target control statement and determining a group of candidate words from the plurality of words includes:

Perform syntactic analysis on the target control statement to determine the core verb of the target control instruction;

Determine each verb in the plurality of words that has the specified syntactic relationship with the core verb as a candidate verb to obtain the set of candidate verbs;

finding a verb having the specified syntactic relationship with any candidate verb in the set of candidate verbs;

When a verb having the specified syntactic relationship with any of the candidate verbs is found, the found verb is added as a candidate verb to the set of candidate verbs.
The method according to claim 2, wherein the associated words of each candidate verb are adjacent words of each candidate verb, and the adjacent words of each candidate verb are located in each candidate verb. The words that follow and are adjacent to each candidate verb;

Determining a target verb from the group of candidate verbs based on the syntactic relationship between each candidate verb in the group of candidate verbs and associated words of each candidate verb includes:

A candidate verb among the group of candidate verbs that satisfies a first filtering condition between adjacent words is determined as the target verb, wherein the first filtering condition includes a syntactic relationship that is the specified syntactic relationship.
The method according to claim 5, wherein determining the candidate verb in the group of candidate verbs that satisfies the first filtering condition between adjacent words as the target verb includes:

The syntactic relationship between any candidate verb in the group of candidate verbs and the adjacent words of any candidate verb is a verb-object relationship, and the part of speech of the adjacent words of any candidate verb is In the case of a noun, any candidate verb is determined as the target verb, wherein the specified syntactic relationship includes the verb-object relationship, and the first filtering condition also includes that the part of speech of the adjacent word is a noun;

In the case where the syntactic relationship between any candidate verb and the adjacent words of the any candidate verb is a parallel relationship, the any candidate verb is determined as the target verb, wherein the specified syntax The relationship includes the parallel relationship.
The method according to claim 2, wherein the associated words of each candidate verb are words that have a syntactic relationship with each candidate verb;

Determining a target verb from the group of candidate verbs based on the syntactic relationship between each candidate verb in the group of candidate verbs and associated words of each candidate verb includes:

A candidate verb in the group of candidate verbs that satisfies a second filtering condition with a word that has a syntactic relationship is determined as the target verb, wherein the second filtering condition includes a syntactic relationship that is the specified syntactic relationship.
The method according to claim 7, wherein determining the candidate verb in the group of candidate verbs that satisfies the second filtering condition with words that have a syntactic relationship as the target verb includes:

The syntactic relationship between any candidate verb in the group of candidate verbs and the words that have a syntactic relationship with the any candidate verb is a verb-object relationship, and the words that have a syntactic relationship with the any candidate verb When the part of speech is a noun, any candidate verb is determined as the target verb, wherein the specified syntactic relationship includes the verb-object relationship, and the second filtering condition also includes the Part of speech is noun;

In the case where the syntactic relationship between any candidate verb and a word having a syntactic relationship with the any candidate verb is a parallel relationship, the any candidate verb is determined as the target verb, wherein, The specified syntactic relationship includes the coordination relationship.
The method according to any one of claims 1 to 8, wherein determining the target word and associated words of the target word as the target event slot of the target control statement includes:

When there are multiple target words, each target word and associated words of each target word are determined as event slots corresponding to each target word;

A combination of event slots corresponding to each target word is determined as the target event slot.
An event slot extraction device includes:

The acquisition unit is configured to acquire the target control statement of the event slot to be extracted, wherein the target control statement includes a plurality of words;

An analysis unit configured to perform syntactic analysis on the target control statement and determine a group of candidate words from the plurality of words, wherein the number of candidate words included in the group of candidate words is less than the number of the plurality of words. the number of words included;

The first determination unit is configured to determine the target word from the group of candidate words based on the syntactic relationship between each candidate word in the group of candidate words and the associated words of each candidate word, wherein , there is a specified syntactic relationship between the target word and the associated words of the target word;

The second determination unit is configured to determine the target word and related words of the target word as the target event slot of the target control statement.
The device according to claim 10, wherein the first determining unit includes:

The first determination module is configured to determine the target verb from the group of candidate verbs based on the syntactic relationship between each candidate verb in the group of candidate verbs and the associated words of each candidate verb, wherein: The set of candidate words includes the set of candidate verbs, and the target word includes the target verb.
The device of claim 11, wherein the analysis unit includes:

An analysis module configured to perform syntactic analysis on the target control statement and determine the core verb of the target control instruction;

The fourth determination module is configured to determine each verb in the plurality of words that has the specified syntactic relationship with the core verb as a candidate verb, and obtain the group of candidate verbs;

A search module configured to search for verbs that have the specified syntactic relationship with any candidate verb in the set of candidate verbs;

An adding module is configured to add the found verb as a candidate verb to the group of candidate verbs when a verb having the specified syntactic relationship with any of the candidate verbs is found.
The device according to claim 11, wherein the associated words of each candidate verb are adjacent words of each candidate verb, and the adjacent words of each candidate verb are located at the location of each candidate verb. words that follow and are adjacent to each candidate verb; the first determination unit includes:

The first determination sub-module is configured to determine the candidate verbs in the group of candidate verbs that satisfy the first filtering condition between adjacent words as the target verb, wherein the first filtering condition includes a syntactic relationship of The specified syntactic relationship.
The device according to claim 13, wherein the first determining sub-module includes:

The first determination subunit is configured such that the syntactic relationship between any candidate verb in the group of candidate verbs and adjacent words to the any candidate verb is a verb-object relationship, and any candidate When the part of speech of the adjacent words of the verb is a noun, any candidate verb is determined as the target verb, wherein the specified syntactic relationship includes the verb-object relationship, and the first filtering condition also includes related The part-of-speech of the adjacent word is a noun;

The second determination subunit is configured to determine any candidate verb as the target when the syntactic relationship between any candidate verb and the adjacent words of the any candidate verb is a parallel relationship. Verb, wherein the specified syntactic relationship includes the parallel relationship.
The device according to claim 11, wherein the first determining module includes:

The second determination sub-module is configured to determine the candidate verbs in the group of candidate verbs that meet the second filtering condition between words with syntactic relationships as the target verb, wherein the second filtering condition includes syntax A relation is the specified syntactic relation.
The device according to claim 15, wherein the second determination sub-module includes:

The third determination subunit is configured such that the syntactic relationship between any candidate verb in the group of candidate verbs and the words that have a syntactic relationship with the any candidate verb is a verb-object relationship, and the syntactic relationship with any candidate verb is a verb-object relationship. When the part-of-speech of a word with a syntactic relationship between a candidate verb is a noun, any candidate verb is determined as the target verb, wherein the specified syntactic relationship includes the verb-object relationship, and the second filtering condition It also includes words with syntactic relationships whose part-of-speech is noun;

The fourth determination subunit is configured to determine the candidate verb as the candidate verb when the syntactic relationship between the candidate verb and the word having a syntactic relationship with the candidate verb is a parallel relationship. Describe the target verb, wherein the specified syntactic relationship includes the parallel relationship.
A computer-readable storage medium includes a stored program, wherein the method of any one of claims 1 to 9 is executed when the program is run.
An electronic device includes a memory and a processor, a computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 9 through the computer program.