Air-ground conversation intelligent semantic understanding method based on multi-task hierarchical network

Abstract

The invention relates to the technical field of air traffic control, in particular to an air-ground conversation intelligent semantic understanding method based on a multi-task hierarchical network, which comprises the following steps: encoding air-ground conversation corpus data to obtain word level representation, sentence level representation and word level mask representation; performing knowledge distillation processing to obtain slot position representation; performing role attention processing and intention attention processing on the sentence level representation, and performing role-intention fusion to obtain role representation and intention representation; performing semantic level bidirectional enhancement interaction processing on the slot position representation, the role representation and the intention representation to obtain word-level features and sentence-level features; a multi-task decoder is adopted to decode the word-level features and the sentence-level features, and a predicted slot position, a predicted role and a predicted intention of the air-ground communication corpus data are obtained; according to the invention, comprehensive and meticulous intelligent semantic understanding of air-ground communication can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of air traffic control, and in particular to an intelligent semantic understanding method for air-land calls based on a multi-task hierarchical network. Background Art

[0002] Air traffic control (ATC) is highly dependent on air-to-ground communications between controllers and pilots, where controllers convey precise situational awareness instructions to pilots via radio, and pilots read back the instructions and operate the aircraft accordingly. However, the rapid growth of air traffic has significantly increased the workload of controllers and pilots, increasing the risk of accidents and incidents caused by misunderstandings in communications between controllers and pilots. With the widespread application of artificial intelligence, the development of an intelligent system that can understand the conversation between controllers and pilots is crucial to assist operators and improve flight safety and efficiency.

[0003] To meet this urgent need, the speech of air-to-ground calls is first converted into text, and then the main purpose and key information of the converted text are extracted and identified to complete semantic understanding. Traditionally, semantic understanding relies on regular expressions and predefined rules to match structured keywords with text instructions. However, the manual templates created by ATC experts are difficult to adapt to the ever-changing application scenarios. Therefore, pre-trained language models (PLMs) have attracted the attention of academia and industry. Various semantic understanding tasks have been explored at different granularities, ranging from word-level named entity recognition and slot filling tasks to sentence-level sentence meaning understanding and intent recognition tasks. In addition, the role detection task, which is also at the sentence level, clarifies the different operational responsibilities of controllers and pilots, ensuring accurate communication and task execution. Although these studies have shown performance improvements, the deep relationship between different granularities has not been fully revealed, especially in terms of the comprehensive utilization of role information. Accurately interpreting the meaning of specialized terms and phrases in utterances is crucial, and discerning the roles and intentions of speakers in different contexts is also an important part of higher-level semantic understanding. However, text representations at different semantic levels of words, syntax, sentences, and concepts in land-air conversations provide insights from different dimensions, and effectively integrating them can enhance understanding. This includes slot filling tasks at the word level, as well as role detection and intent recognition tasks at the sentence level, which are inherently heterogeneous because slot filling involves mapping each word or phrase in a sentence to a specific semantic label, so it is at the word level; role detection distinguishes the identities of different speakers by identifying the word order structure of the sentence, so it is at the syntactic level; intent recognition captures higher-level semantic intent by analyzing the entire sentence and its key actions to determine the purpose or meaning behind the discourse, so it is at the sentence and concept level. Therefore, how to mine and integrate task relationships at different granularities is the main issue. Summary of the invention

[0004] In view of the above problems, the present invention provides an intelligent semantic understanding method for land-to-air calls based on a multi-task hierarchical network, which solves the technical problem that the existing technology fails to mine and integrate tasks of different granularities in the semantic recognition task of land-to-air calls.

[0005] The present invention provides an intelligent semantic understanding method for land-to-air conversation based on a multi-task hierarchical network, wherein the multi-task hierarchical network processes land-to-air conversation corpus data including the following steps:

[0006] Step S1, encoding the land-air communication corpus data at the word level and sentence level to obtain word level representation and sentence level representation;

[0007] Step S2, performing word-level mask encoding on the land-air communication corpus data to obtain a word-level mask representation, performing knowledge distillation processing based on the word-level representation and the word-level mask representation to obtain a slot representation;

[0008] Step S3, performing role attention processing and intention attention processing on the sentence level representation respectively, and then performing role-intention fusion to obtain role representation and intention representation;

[0009] Step S4, performing semantic-level bidirectional enhancement interactive processing on the slot representation, role representation, and intention representation to obtain word-level features and sentence-level features;

[0010] Step S5: Use a multi-task decoder to decode the word-level features and sentence-level features to obtain the predicted slot, predicted role, and predicted intention of the land-air communication corpus data.

[0011] Preferably, the step S1 specifically includes:

[0012] Step S1-1, encoding the land-air communication corpus data through a shared encoder to obtain a shared representation; the shared encoder is a BERT model;

[0013] Step S1-2, encoding the shared representation through a feature-specific encoder to obtain a word-level representation and a sentence-level representation; the feature-specific encoder is a bidirectional LSTM network.

[0014] Preferably, the step S1-1 specifically includes: the sentences in the text of the land-air conversation content form an input sentence word sequence x, where x={x1, x2, ..., x n}, where x n is the nth word of the input sentence; the BERT model is used to generate a shared representation, denoted as E = {e [CLS] , e1, e2, ..., en , e [SEP]}, where e n is the shared representation corresponding to the nth word in x, e [CLS] and e [SEP] are respectively sentence boundaries; the step S1-2 specifically comprises: inputting the shared representation E into the first bidirectional LSTM network and the second bidirectional LSTM network respectively, and obtaining the word level representation H w and sentence level representation H u .

[0015] Preferably, the step S2 specifically includes:

[0016] Step S2-1, establishing an external knowledge base based on standard terminology of air traffic control scenarios; establishing an internal knowledge base based on historical land-air communication corpus data; and combining the internal knowledge base and the external knowledge base to form a candidate mask word set;

[0017] Step S2-2, after processing the land-air communication corpus data using the masked language model and the shared encoder, input the masked feature-specific encoder for encoding to obtain a word-level masked representation;

[0018] Step S2-3: perform knowledge distillation processing on the word level representation and the word level mask representation based on the teacher model and the student model to obtain a slot representation.

[0019] Preferably, in step S2-1, the specific steps of establishing the internal knowledge base based on the historical land-air communication corpus data include: using TF-IDF word frequency processing technology to divide the text in the historical land-air communication corpus data into words, and sorting the divided words according to the frequency of word occurrence, and selecting a part of the words with the highest ranking from the sorted words as the internal knowledge base;

[0020] Step S2-2 specifically includes: using a masked language model to process the land-air communication corpus data to obtain a masked word sequence, x′={x1, [MASK], ..., x n}, [MASK] represents the replaced mask mark, and the shared encoder is used to process x′ to obtain E′, E′={e′ [CLS] , e′1, e′ [MASK］ , ..., e′ n , e′ [SEP]}, where e′ n is the shared representation corresponding to the nth word in x′, e′ [MASK] is the shared representation corresponding to the mask mark, e′ [CLS] and e′ [SEP]are masked sentence boundaries respectively; E′ is calculated in the third bidirectional LSTM network to obtain the word-level mask representation H m .

[0021] Step S2-3 specifically includes: the teacher model converts H w With H m Splice and process to generate the teacher representation F t The student model for H w Processing is performed to generate slot representation F s , the expression is:

[0022]

[0023] F s =Student(H w )

[0024] in, represents the concatenation operation, Teacher(·) represents the teacher model, and Student(·) represents the student model.

[0025] Preferably, the step S3 specifically includes:

[0026] Step S3-1, performing calculation based on the sentence level representation to obtain a role attention weight and an intention attention weight; based on the role attention weight and the intention attention weight, obtaining a role context and an intention context respectively;

[0027] Step S3-2: Use the role-intention fusion layer to project the role context into the intent latent space to obtain a fused projection, concatenate the intent context and the fused projection to obtain a role representation; map the role representation to the intent label space to obtain an intent representation.

[0028] Preferably, the step S3-1 specifically includes: calculating the character attention weight A by the following formula: r and the intention attention weight A i :

[0029] A r =softmax(W r H u +b r )

[0030] A i =softmax(W i H u +b i )

[0031] Among them, W r and b rRepresent the weight and bias of the character's attention respectively. The softmax function normalizes the calculation result to the character's attention weight A. r , W i and b i Represent the weight and bias of the intention attention respectively;

[0032] By adding H u With transposed attention weights and Multiply them to get the role context C r and intent context C i :

[0033]

[0034]

[0035] in, and Denote the generated weighted role and intention representations, V r and V i are the dimension conversion matrices of roles and intentions respectively;

[0036] The step S3-2 specifically includes: using the role-intention fusion layer to calculate and obtain the fusion projection C f : The expression is:

[0037] C f =fusion(C r )

[0038] Where fusion represents the role-intention fusion layer;

[0039] Character representation F r and the intention to express F i The expression is:

[0040]

[0041] F i =W u F r +b u

[0042] Among them, W u and b u Represent the weight and bias of the sentence respectively.

[0043] Preferably, the step S4 specifically includes:

[0044] The slot representation, role representation and intention representation are sequentially determined as the semantic level sequence of the semantic level bidirectional enhanced interaction processing, that is, S1=Fs , S2=F r , S3=F i , S l is the input representation of the lth layer;

[0045] The interaction score is calculated within the hierarchy using a bidirectional bilinear transformation, expressed as:

[0046]

[0047]

[0048] in, is the reverse interaction score of the lth layer, is the forward interaction score of the lth layer, Score(·,·) represents the interaction score between the calculated layers, is the transpose of the input representation of the lth layer, is the reverse score matrix, is the forward score matrix;

[0049] Calculate the backward and forward pass vectors for layer l and The expression is:

[0050]

[0051]

[0052] in, and For the backward and forward pass weights, and Bias for reverse and forward transfer;

[0053] By combining and and and Recursively aggregate the calculations to obtain word-level features and sentence-level features, expressed as:

[0054]

[0055]

[0056] in, represents the aggregated word-level features from layer L to layer l, where L represents the total number of layers. represents the aggregated sentence-level features from level 1 to level l;

[0057] Finally, and They are respectively used as the word-level features and sentence-level features.

[0058] Preferably, the step S5 specifically includes:

[0059] Step S5-1: Input to the slot classifier, and then pass through a CRF layer to obtain the predicted slot;

[0060] Step S5-2: With F r Concatenate and input into the role classifier to obtain the predicted role;

[0061] Step S5-3: F i Input to the intent quantity classifier to get the predicted number of intents. Input to the intent classifier to obtain the intent probability, and select the top c with the highest probability from the predicted number of intents * The intention is used as the predicted intention.

[0062] Preferably, by training the multi-task hierarchical network, with the goal of minimizing the joint target training loss, back propagation is performed to update the model parameters, and after obtaining the optimal parameters of the model, the trained multi-task hierarchical network is used for the intelligent semantic understanding of land-air calls; the joint target training loss of the multi-task hierarchical network is minimized. The expression is:

[0063]

[0064] in, is the slot filling loss, Detection loss for the character, To identify the loss of intent, Slot distillation loss, λ, μ and η are loss weights;

[0065]

[0066] in, is the true label of the kth slot of the ith word, is the predicted slot probability of the kth slot of the ith word, n is the total number of words, and K is the total number of slot labels;

[0067]

[0068] in, and are the true binary labels for the controller and pilot roles, respectively; and are the predicted probabilities for the controller and pilot roles, respectively;

[0069]

[0070] in, is the Kronecker delta function, Indicates the probability that the number of intentions is predicted to be c. When it matches the actual intent number c is 1, otherwise is 0, represents the binary true label of the mth intent, is the predicted probability of the mth intention, M is the total number of intentions;

[0071]

[0072]

[0073] in, is the predicted probability distribution of the i-th word, m i Indicates whether the i-th word is masked, z i is the unique hot representation of the i-th word;

[0074]

[0075] in, and are the probability distributions of teacher representation and student representation, respectively. express and The KL divergence of is a finite set space, χ is The elements in the collection, They represent the target probability and approximate probability corresponding to x respectively.

[0076] Compared with the prior art, the present invention has at least the following beneficial effects:

[0077] (1) The present invention generates features of words in sentences by inputting the ground-to-air conversation text into a corpus encoder composed of a shared encoder and a feature-specific Bi-LSTM encoder, and can simultaneously perform slot filling, role detection and intent recognition, thereby achieving a comprehensive understanding of the conversation between the controller and the pilot and improving the efficiency and accuracy of the overall semantic understanding.

[0078] (2) The present invention introduces word-level knowledge mask slot distillation, utilizes knowledge-guided MLM, guides slot representation through teacher-student distillation, builds a special ATC knowledge base, focuses on keywords and enhances the generalization ability of the model, thereby improving the model's ability to recognize key terms and enhancing its adaptability to different scenarios.

[0079] (3) The present invention adopts a sentence-level role-aware intention attention mechanism to extract label space features and integrate roles into the intention space. Subsequently, through the bidirectional enhanced interaction of semantic levels, information is propagated and aggregated between multiple semantic understanding tasks, which can effectively distinguish and integrate semantic levels of different granularities, improve the recognition accuracy of slot filling, role detection and intent recognition, and because of the recursive formula structure, the framework has good scalability. The current model supports four levels of semantic levels, corresponding to three tasks. On this basis, more levels and tasks can be added without modifying the existing module framework. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] The drawings are only for the purpose of illustrating particular embodiments and are not to be construed as limiting the invention.

[0081] Figure 1 The overall framework diagram of the intelligent semantic understanding method for land-to-air calls based on a multi-task hierarchical network provided by the present invention.

[0082] Figure 2 A schematic diagram of the knowledge shielding slot distillation construction process provided by the present invention.

[0083] Figure 3 Schematic diagram of the character-aware intention attention mechanism provided for the present invention.

[0084] Figure 4 A schematic diagram of the semantic level bidirectional enhanced interaction provided by the present invention.

[0085] Figure 5 A flow chart of the method for intelligent semantic understanding of land-to-air calls based on a multi-task hierarchical network provided by the present invention. DETAILED DESCRIPTION

[0086] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. In addition, the present invention can also be implemented in other ways different from those described herein, and therefore, the protection scope of the present invention is not limited by the specific embodiments disclosed below.

[0087] This paper proposes an intelligent semantic understanding method for air-to-land calls based on a multi-task hierarchical network, and introduces word-level knowledge mask slot distillation to improve the sensitivity and generalization ability of the word level. This goal is achieved by using a masked language model and knowledge distillation guided by ATC knowledge; at the same time, a sentence-level role-aware intent attention mechanism is involved, and two attention mechanisms are used to extract and explicitly fuse role and intent features. The attention mechanism minimizes semantic ambiguity by mapping the same corpus to the role and intent label space, and the fusion layer uses role information as context to enrich the intent representation; finally, the semantic level is bidirectionally enhanced to capture the relationship between the hierarchical semantic representations. Specifically learn the bidirectional interaction between the levels and recursively generate the final multi-task representation.

[0088] In order to illustrate the effectiveness of the method proposed by the present invention, the above technical solution of the present invention is described in detail below through a specific embodiment. Figure 1 , Figure 5 As shown, a method for intelligent semantic understanding of land-air calls based on a multi-task hierarchical network is disclosed, and the specific implementation steps are as follows:

[0089] Step S1, encoding the land-air communication corpus data at the word level and sentence level to obtain word level representation and sentence level representation;

[0090] The content of the ground-to-air communication data in the embodiment of the present invention is derived from the ground-to-air communication content between the controller and the pilot in air traffic control, and the ground-to-air communication content includes the speech of the controller and the speech of the pilot. The original audio of the speech of the controller and the speech of the pilot is transcribed into text to obtain the text of multiple segments of the ground-to-air communication content as the ground-to-air communication data.

[0091] For the encoding process of the land-air communication corpus data, the present invention successively encodes the corpus data through a shared encoder and a feature-specific encoder to obtain a word-level representation and a sentence-level representation. The specific process is as follows:

[0092] (1) Shared encoder

[0093] The shared encoder of the present invention adopts the BERT model which is a widely accepted pre-trained language model (PLM). The BERT model is obtained by pre-training on a large-scale corpus. The sentences in the text of the land-air call content form an input sentence word sequence x, where x = {x1, x2, ..., x n}, where x n is the nth word of the input sentence. Using the knowledge of the BERT model pre-trained on a large-scale corpus, the generated shared representation is denoted as E = {e [CLS] , e1, e2, ..., e n , e[SEP]}, where e n is the shared representation corresponding to the nth word in x, e [CLS] and e [SEP] Statement boundaries added at the beginning and end of a statement, respectively.

[0094] (2) Feature-specific encoder

[0095] The feature-specific encoder of the present invention is a bidirectional LSTM (Bi-LSTM) network, which inputs the shared representation E into two bidirectional LSTM (Bi-LSTM) networks of specific features to obtain word-level representations H w and sentence level representation H u .

[0096] H w =Bi-LSTM w (E) (1)

[0097] H u =Bi-LSTM u (E) (2)

[0098] Among them, Bi-LSTM w (·) is a word-level bidirectional LSTM network, Bi-LSTM u (·) is a sentence-level bidirectional LSTM network.

[0099] Step S2, performing word-level mask encoding on the land-air communication corpus data to obtain a word-level mask representation, performing knowledge distillation processing based on the word-level representation and the word-level mask representation to obtain a slot representation;

[0100] In air traffic control scenarios, accurate identification of keywords is crucial to ensure the effectiveness and clarity of air-to-ground communication, which is called the slot filling task. Specifically, when processing an air-to-ground communication instruction from a pilot or controller, the slot filling task requires the assignment of a slot label to each word in the sentence. Therefore, the present invention provides a method for processing word-level knowledge mask slot distillation, such as Figure 2 shown.

[0101] First, for the land-air communication corpus data, the present invention provides a knowledge-guided masked language model (MLM) for masking specific keywords and predicting their content, thereby enhancing the model's sensitivity to these keywords while reducing interference from irrelevant words. Specifically, the present invention uses TF-IDF word frequency processing technology to divide the text in the historical land-air communication corpus data into words, and sorts the divided words according to the frequency of word occurrence, and selects a part of the words with the highest ranking from the sorted words as the internal knowledge base. In addition, based on the standard terminology of air traffic control scenarios, an external knowledge base is established, and the external knowledge base includes letters, numbers, and abbreviations of air traffic control terms.

[0102] The internal and external knowledge together constitute a candidate mask word set. The mask language model is used to process the original land-air communication corpus data to generate an input sentence mask word sequence x′, x′={x1, [MASK], ..., x n}, where some of the original words are replaced with [MASK] tags, [MASK] represents the replaced mask tag, and the shared representation corresponding to x′ is recorded as E′, E′={e′ [CLS] , e′1, e′ [MASK] , ..., e′ n , e′ [SEP]}, where e′ n is the shared representation corresponding to the nth word in x′, e′ [MASK] is the shared representation corresponding to the mask mark, e′ [CLS] and e′ [SEP] The masked sentence boundaries are added at the beginning and end of the sentence represented by the mask, respectively.

[0103] Use another bidirectional LSTM network to calculate E′ and obtain the word-level mask representation H m , as shown below:

[0104] H m =Bi-LSTM m (E′) (3)

[0105] Among them, Bi-LSTM m (·) is a bidirectional LSTM network with word-level masking.

[0106] The loss function of the knowledge-guided masked language model The expression is:

[0107]

[0108] in, is the predicted probability distribution of the i-th word, mi Indicates whether the i-th word is masked, z i is the unique hot representation of the i-th word. is obtained by applying weights w and bias b to the encoder features e′ i The predicted probability distribution of the i-th word is obtained by normalizing it with the softmax function. If the word is masked, then m i =1, otherwise 0. Loss function of masked language model is the true distribution z at the mask position i With the predicted distribution By minimizing this loss function, the model is trained to accurately predict the masked words based on the context.

[0109] After obtaining the word-level mask representation, the present invention provides a knowledge distillation method for subsequent processing. Specifically, the knowledge distillation method is based on a teacher model and a student model. The teacher model converts H w With H m Splice and process to generate the teacher representation F t The student model for H w Processing is performed to generate slot representation F s The expression is:

[0110]

[0111] F s =Student(H w ) (6)

[0112] in, represents the concatenation operation, Teacher(·) represents the teacher model, and Student(·) represents the student model. In order to align the representation and promote the effective distillation from the teacher model to the student model, F t and F s Unify the dimensions. In addition, use softmax to t and F s Normalize it and get its probability distribution and

[0113] The goal of the knowledge distillation method based on the teacher model and the student model is to minimize the target distribution With approximate distribution The Kullback-Leibler (KL) divergence between the two, the objective function of the knowledge distillation method The expression is:

[0114]

[0115] in, express and KL divergence, χ is a finite set space, χ is an element in the χ set, They represent the target probability and approximate probability corresponding to χ respectively.

[0116] Finally, the overall tank distillation loss in this step The expression is:

[0117]

[0118] Step S3: perform role attention processing and intention attention processing on the sentence level representation respectively, and then perform role-intention fusion to obtain role representation and intention representation.

[0119] In air-to-ground conversations, the speeches of controllers and pilots usually have different semantic functions. Controllers mainly issue operational instructions, while pilots confirm or respond to these instructions. Therefore, the role detection task is an important task in the intelligent semantic understanding task of air-to-ground conversations, which is used to identify and distinguish the identities of different speakers (controllers, pilots). The intention recognition task analyzes the dialogue input of pilots and controllers to determine the operations they expect to perform or the important information they convey. In addition, different roles show different tendencies for certain intentions. For example, controllers often issue instruction intentions such as "altitude intentions", while pilots are more inclined to "request intentions" and "status report intentions".

[0120] The present invention provides a sentence-level role-aware intention attention mechanism, which integrates the contextual relevance of the role with its corresponding intention, and can simultaneously complete the role detection task and the intention recognition task, such as Figure 3 By leveraging role information, we aim to achieve more accurate intent prediction.

[0121] Specifically, the present invention utilizes two independent attention mechanisms to represent H from the same sentence level. u The role representation and intent representation are extracted from the , respectively, to ensure a clear distinction between them. For the role attention processing, the role attention weight A r The calculation expression is:

[0122] A r =softmax(W r H u +b r ) (9)

[0123] Where W r and b rRepresent the weight and bias of the character's attention respectively. The softmax function normalizes the calculation result to the character's attention weight A. r .

[0124] Similarly, the intention attention weight A i The calculation expression is:

[0125] A i =softmax(W i H u +b i ) (10)

[0126] Where W i and b i They represent the weight and bias of the intention attention respectively.

[0127] Then, by adding H u With transposed attention weights and Multiply them to get the role context C r and intent context C i :

[0128]

[0129]

[0130] in, and Denote the generated weighted role and intention representations, V r and V i The dimension conversion matrices are role and intent respectively.

[0131] In order to effectively fuse the role and intent information, a role-intent fusion layer is used to transform the role context C r Projected to the intention latent space, the fused projection C is obtained f , the expression of the role-intention fusion layer calculation process is:

[0132] C f =fusion(C r ) (13)

[0133] Where fusion represents the role-intention fusion layer. The role-intention fusion layer converts C r Projection to C f , and activate each role through the sigmoid function.

[0134] Finally, to ensure that the role and intent representations are tightly coupled and better adapt to the sentence-based feature distribution, we define It represents the role.i It is obtained by the following expression:

[0135] F i =W u F r +b u (14)

[0136] Among them, W u and b u Represent the weight and bias of the sentence respectively. Through such calculation, F r is mapped to the intent label space to obtain F i .

[0137] Step S4: perform semantic-level bidirectional enhanced interactive processing on the slot representation, role representation, and intent representation to obtain word-level and sentence-level features.

[0138] Figure 4 The process of the semantic level bidirectional enhancement interactive processing provided by the present invention is shown. The input includes representations S1, S2, S3 at multiple different semantic levels, and the final output is the aggregated word-level features. and sentence-level features

[0139] The present invention provides a semantic hierarchical bidirectional enhanced interaction design, which explores the semantic relationship between multiple tasks by using bidirectional interaction scores. Through the bidirectional propagation layer, the present invention combines the bidirectional interaction in hierarchical semantics, thereby finally forming aggregated word-level and sentence-level features. Semantic hierarchical sequence S = (S1, ..., S L ) as input, where L represents the total number of levels. In the practical application of the present invention, the semantic level sequence is defined as: S = (F s , F r , F i ), corresponding to slot, role, and intent representation respectively. This semantic hierarchy sequence provides a detailed and comprehensive semantic representation.

[0140] The hierarchical semantics of S has bidirectional interdependence, and the information flow from fine-grained semantic details to the overall conceptual meaning is different from the information flow from the overall conceptual meaning to fine-grained semantic details. Therefore, the present invention calculates the interaction score within the hierarchy by using a bidirectional bilinear transformation, and the expression is:

[0141]

[0142] in, is the reverse interaction score of the lth layer, is the forward interaction score of the lth layer, Score(·,·) represents the interaction score between the calculated layers, S l is the input representation of the lth layer, is the transpose of the input representation of the lth layer, is the reverse score matrix, is the forward score matrix.

[0143] Subsequently, the present invention applies backward and forward semantic transfer at each level to calculate the backward and forward transfer vectors of the lth layer and The expression is:

[0144]

[0145] in, and For the backward and forward pass weights, and Transfer bias for reverse and forward directions.

[0146] Next, by combining the transfer vector and Interaction score and Recursively calculate the aggregated word-level and sentence-level features. The aggregated word-level features are obtained as follows:

[0147]

[0148] in, represents the aggregated word-level features from layer L to layer l, where L represents the total number of layers.

[0149] Similarly, the aggregated sentence-level features are obtained as follows:

[0150]

[0151] in, represents the aggregated sentence-level features from level 1 to level l.

[0152] final, and As the final aggregated word-level and sentence-level features respectively. In the practical application of the present invention, L = 3, that is, using and Specifically, we gradually transfer from intent features to role and slot features, and finally form word-level features that integrate sentence features. Starting from the slot features, the role and intent features are gradually integrated, and finally the sentence-level features that integrate the word features are formed.

[0153] Step S5: Use a multi-task decoder to decode the word-level and sentence-level features to obtain predicted slots, predicted roles, and predicted intents.

[0154] The multi-task decoder includes decoders for the following three tasks:

[0155] (1) Slot decoder:

[0156] is fed into a slot classifier and then passed through a conditional random field (CRF) layer to obtain the predicted slot probability The CRF layer determines the most likely sequence of slot labels by considering individual slot probabilities and transitions between consecutive slots. The loss function for the slot filling loss is the cross entropy loss

[0157]

[0158] in, is the true label of the kth slot of the ith word, is the predicted slot probability of the kth slot of the uth word, n is the total number of words, and K is the total number of slot labels.

[0159] (2) Role decoder:

[0160] The present invention will With F r Concatenate and input into the role classifier to obtain the predicted role probability The loss function of character detection loss is defined as follows:

[0161]

[0162] in, and are the true binary labels of the controller and pilot roles respectively. The predicted probabilities of the controller and pilot roles are expressed as and And their relationship is

[0163] (3) Intent decoder:

[0164] The present invention intends to represent F i Input to the intent number classifier, which normalizes each possible intent number and outputs the predicted number of intents in the sentence Where C represents the maximum number of intents considered.

[0165] The present invention will Input to the intent classifier to obtain the intent probability. Select the top c with the highest probability * Multi-label intent set

[0166] Intent Identification Loss Includes binary cross entropy loss for multi-label intent recognition and cross entropy loss for determining the number of intents:

[0167]

[0168] in, is the Kronecker delta function, when When it matches the actual intent number c is 1, otherwise is 0, represents the probability that the number of intentions is predicted to be c, represents the binary true label of the mth intent, is the predicted probability of the mth intent, and M is the total number of intents.

[0169] Joint objective training loss of the model of the present invention Loss for slot filling Character Detection Loss Intent Identification Loss And tank distillation loss The weighted sum of , where λ, μ, and η are loss weights:

[0170]

[0171] Through the learning of the loss function during the training process, the model parameters are updated through back propagation to obtain the optimal parameters of the model, so that the above multiple tasks can be well represented. After the training, the model is used as a language model in the field of air traffic control and can be used for comprehensive and in-depth intelligent semantic understanding of air-to-ground calls.

[0172] The present invention discloses a method for intelligent semantic understanding of air-to-ground communication based on a multi-task hierarchical network. Aiming at the problem that frequent air-to-ground communication in air traffic control causes heavy load on controllers and pilots, thus leading to communication misunderstandings, and easily causing accident symptoms or even accidents, a novel end-to-end multi-task hierarchical network is proposed. Through a multi-task framework, comprehensive and accurate semantic understanding is intelligently achieved. The slot filling, role detection and intention recognition tasks are integrated to achieve comprehensive semantic intelligent understanding of different levels of words, syntax, sentences and concepts in air-to-ground communication, ensuring clear and consistent communication between controllers and pilots, reducing misunderstandings, and improving safety and operational efficiency. Compared with the prior art, the present invention can effectively achieve comprehensive and detailed intelligent semantic understanding of air-to-ground communication by effectively processing multiple tasks, ensuring flight safety and improving flight efficiency.

[0173] Although the specific embodiments of the present invention depict various actions or steps in a specific order, this should be understood as requiring such actions or steps to be performed in the specific order shown or in a sequential order, or requiring all illustrated actions or steps to be performed to obtain the desired results. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although several specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Certain features described in the context of a separate embodiment can also be implemented in a single implementation in combination. On the contrary, the various features described in the context of a single implementation can also be implemented in multiple implementations individually or in any suitable sub-combination. The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or replacements that can be easily thought of by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

[0174] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for intelligent semantic understanding of land-air calls based on a multi-task hierarchical network, characterized in that: The processing of the land-air communication data by the multi-task hierarchical network comprises the following steps: Step S1, encoding the land-air communication corpus data at the word level and sentence level to obtain word level representation and sentence level representation; Step S2, performing word-level mask encoding on the land-air communication corpus data to obtain a word-level mask representation, performing knowledge distillation processing based on the word-level representation and the word-level mask representation to obtain a slot representation; Step S3, performing role attention processing and intention attention processing on the sentence level representation respectively, and then performing role-intention fusion to obtain role representation and intention representation; Step S4, performing semantic-level bidirectional enhancement interactive processing on the slot representation, role representation, and intention representation to obtain word-level features and sentence-level features; Step S5: Use a multi-task decoder to decode the word-level features and sentence-level features to obtain the predicted slot, predicted role, and predicted intention of the land-air communication corpus data.

2. The intelligent semantic understanding method for air-land calls based on a multi-task hierarchical network according to claim 1 is characterized in that: The step S1 specifically includes: Step S1-1, encoding the land-air communication corpus data through a shared encoder to obtain a shared representation; the shared encoder is a BERT model; Step S1-2, encoding the shared representation through a feature-specific encoder to obtain a word-level representation and a sentence-level representation; the feature-specific encoder is a bidirectional LSTM network.

3. The method for intelligent semantic understanding of land-air calls based on a multi-task hierarchical network according to claim 2 is characterized by: The step S1-1 specifically includes: forming an input sentence word sequence x from the text of the land-air conversation content, where x={x1, x2, ..., x n }, where x n is the nth word of the input sentence; the BERT model is used to generate a shared representation, denoted as E = {e [CLS] , e1, e2, ..., e n , e [SEP] }, where e n is the shared representation corresponding to the nth word in x, e [CLS] and e [SEP] They are statement boundaries; The step S1-2 specifically includes: inputting the shared representation E into the first bidirectional LSTM network and the second bidirectional LSTM network respectively to obtain the word level representation H w and sentence level representation H u .

4. The method for intelligent semantic understanding of air-to-ground calls based on a multi-task hierarchical network according to claim 3 is characterized in that: The step S2 specifically includes: Step S2-1, establishing an external knowledge base based on standard terminology of air traffic control scenarios; establishing an internal knowledge base based on historical land-air communication corpus data; and combining the internal knowledge base and the external knowledge base to form a candidate mask word set; Step S2-2, after processing the land-air communication corpus data using the masked language model and the shared encoder, input the masked feature-specific encoder for encoding to obtain a word-level masked representation; Step S2-3: perform knowledge distillation processing on the word level representation and the word level mask representation based on the teacher model and the student model to obtain a slot representation.

5. The method for intelligent semantic understanding of land-air calls based on a multi-task hierarchical network according to claim 4 is characterized by: In step S2-1, the specific steps of establishing the internal knowledge base based on the historical land-air communication corpus data include: using TF-IDF word frequency processing technology to divide the text in the historical land-air communication corpus data into words, and sorting the divided words according to the frequency of word occurrence, and selecting a part of the words with the highest ranking from the sorted words as the internal knowledge base; Step S2-2 specifically includes: using a masked language model to process the land-air communication corpus data to obtain a masked word sequence, x′={x1, [MASK], ..., x n }, [MASK] represents the mask mark of the replacement, and the shared encoder is used to process x′ to obtain E′, E′={e′ [CLS] , e′1, e′ [MASK] , ..., e′ n , e′ [SEP] }, where e′ n is the shared representation corresponding to the nth word in x′, e′ [MASK] is the shared representation corresponding to the mask mark, e′ [CLS] and e′ [SEP] are masked sentence boundaries respectively; E′ is calculated in the third bidirectional LSTM network to obtain the word-level mask representation H m ; Step S2-3 specifically includes: the teacher model converts H w With H m Splice and process to generate the teacher representation F t , the student model for H w Processing is performed to generate slot representation F s , the expression is: F s =Student(H w ) in, represents the concatenation operation, Teacher(·) represents the teacher model, and Student(·) represents the student model.

6. The method for intelligent semantic understanding of air-to-ground calls based on a multi-task hierarchical network according to claim 5 is characterized in that: The step S3 specifically includes: Step S3-1, performing calculation based on the sentence level representation to obtain a role attention weight and an intention attention weight; based on the role attention weight and the intention attention weight, obtaining a role context and an intention context respectively; Step S3-2: Use the role-intention fusion layer to project the role context into the intent latent space to obtain a fused projection, concatenate the intent context and the fused projection to obtain a role representation; map the role representation to the intent label space to obtain an intent representation.

7. The intelligent semantic understanding method for air-to-ground calls based on a multi-task hierarchical network according to claim 6 is characterized by: The step S3-1 specifically includes: calculating the character attention weight A by the following formula: r and the intention attention weight A i : A r =softmax(W r H u +b r ) A i =softmax(W i H u +b i ) Among them, W r and b r Represent the weight and bias of the character's attention respectively. The softmax function normalizes the calculation result to the character's attention weight A. r , W i and b i Represent the weight and bias of the intention attention respectively; By adding H u With transposed attention weights and Multiply them to get the role context C r and intent context C i : in, and Denote the generated weighted role and intention representations, V r and V i are the dimension conversion matrices of roles and intentions respectively; The step S3-2 specifically includes: using the role-intention fusion layer to calculate and obtain the fusion projection C f : The expression is: C f =fusion(C r ) Where fusion represents the role-intention fusion layer; Character representation F r and the intention to express F i The expression is: F i =W u F r +b u Among them, W u and b u Represent the weight and bias of the sentence respectively.

8. The method for intelligent semantic understanding of air-to-ground calls based on a multi-task hierarchical network according to claim 7 is characterized in that: The step S4 specifically includes: The slot representation, role representation and intention representation are sequentially determined as the semantic level sequence of the semantic level bidirectional enhanced interaction processing, that is, S1=F s , S2=F r , S3=F i , S l is the input representation of the lth layer; The interaction score is calculated within the hierarchy using a bidirectional bilinear transformation, expressed as: in, is the reverse interaction score of the lth layer, is the forward interaction score of the lth layer, Score(·,·) represents the interaction score between the calculated layers, is the transpose of the input representation of the lth layer, is the reverse score matrix, is the forward score matrix; Calculate the backward and forward pass vectors for layer l and The expression is: in, and For the backward and forward pass weights, and Bias for reverse and forward transfer; By combining and and and Recursively aggregate the calculations to obtain word-level features and sentence-level features, expressed as: in, represents the aggregated word-level features from layer L to layer l, where L represents the total number of layers. represents the aggregated sentence-level features from level 1 to level l; Finally, and They are respectively used as the word-level features and sentence-level features.

9. The method for intelligent semantic understanding of air-to-ground calls based on a multi-task hierarchical network according to claim 8 is characterized in that: The step S5 specifically includes: Step S5-1: Input to the slot classifier, and then pass through a CRF layer to obtain the predicted slot; Step S5-2: With F r Concatenate and input into the role classifier to obtain the predicted role; Step S5-3: F i Input to the intent quantity classifier to get the predicted number of intents. Input to the intent classifier to obtain the intent probability, and select the top c with the highest probability from the predicted number of intents * The intention is used as the predicted intention.

10. The intelligent semantic understanding method for air-to-ground calls based on a multi-task hierarchical network according to claim 9 is characterized by: By training the multi-task hierarchical network, with the goal of minimizing the joint target training loss, back propagation is performed to update the model parameters, and after obtaining the optimal parameters of the model, the trained multi-task hierarchical network is used for the intelligent semantic understanding of land-air calls; the joint target training loss of the multi-task hierarchical network is minimized. The expression is: in, is the slot filling loss, Detection loss for the character, For the purpose of identifying losses, Slot distillation loss, λ, μ and η are loss weights; in, is the true label of the kth slot of the ith word, is the predicted slot probability of the kth slot of the ith word, n is the total number of words, and K is the total number of slot labels; in, and are the true binary labels for the controller and pilot roles, respectively; and are the predicted probabilities for the controller and pilot roles, respectively; in, is the Kronecker delta function, Indicates the probability that the number of intentions is predicted to be c. When it matches the actual intent number c is 1, otherwise is 0, represents the binary true label of the mth intent, is the predicted probability of the mth intention, M is the total number of intentions; in, is the predicted probability distribution of the i-th word, m i Indicates whether the i-th word is masked, z i is the unique hot representation of the i-th word; in, and are the probability distributions of teacher representation and student representation, respectively. express and The KL divergence of is a finite set space, x is The elements in the collection, They represent the target probability and approximate probability corresponding to x respectively.

Images