Information processing method and device, storage medium and electronic equipment

[0028] The technical solutions in the embodiments of the present application will be described clearly and completely in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative work shall fall within the protection scope of this application.

[0029] Big data refers to a collection of data that cannot be captured, managed, and processed with conventional software tools within a certain time frame. It is a massive amount of data that requires a new processing model to have stronger decision-making power, insight discovery and process optimization capabilities , High growth rate and diversified information assets. With the advent of the cloud era, big data has also attracted more and more attention. Big data requires special technology to effectively process a large amount of data that can be tolerated over time. Cloud technology is the processing technology suitable for big data.

[0030] Among them, cloud technology refers to a hosting technology that unifies a series of resources such as hardware, software, and networks in a wide area network or a local area network to realize data calculation, storage, processing, and sharing. In actual applications, cloud technology can be a general term for cloud computing platform-based network technology, information technology, integration technology, management platform technology, application technology, etc. It can form a resource pool, which can be used as needed, which is flexible and convenient.

[0031] Artificial Intelligence (AI) is a theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, acquire knowledge and use knowledge to obtain the best results, so that machines have perception , Reasoning and decision-making functions. In this application, artificial intelligence and cloud technology are combined to achieve effective processing of user trajectory big data.

[0032] There are two main problems in the implementation of the prior art. The first is that the spatial-temporal correlation within the trajectory is not sufficiently modeled. When the trajectory is continuously missing, the position before and after the missing moment has weak spatial constraints on the time to be interpolated. It is difficult to achieve the desired result. Another problem is that the long movement history of each user is ignored. Therefore, if the mobile data is very sparse (for example, less than 5 points per day), it is impossible to recover the lost locations with fine time granularity (for example, every 30 minutes). Based on this, this application provides figure 1 The schematic diagram of the system architecture shown here aims to recover the user's overall mobile data from the sparse records by considering the user's historical movement patterns in a given time interval.

[0033] In this application, trajectory enhancement is regarded as a text enhancement problem. In this system, since the learned "corpus" is a massive trajectory, its basic "dictionary" is composed of trajectory points. First, there is no directly usable word vector. Secondly, with the refined needs of the business, after the grid scale is as fine as 100 meters, the grid of a city will reach hundreds of thousands of scales. Therefore, the system first builds a pre-training module to realize the vector representation of the dot trace. The module extracts and embeds the user’s features. The features include but are not limited to the latitude and longitude of the dot trace, the serial number of the dot trace in the entire trajectory, and the trajectory date. , User desensitization ID, etc. In specific implementation, these features are transformed into a fixed-length vector with a dimension of [50,100] through a fully connected transformation neural network.

[0034] For the pre-training module, it mainly provides the function of reducing the input scale, so as to ensure that the system can operate stably as the trajectory data increases, and to ensure the robustness and scalability of the system. In the pre-training module, important POI (Point of Interest) data can be introduced to moderately reduce the dimensionality of the "dictionary" composed of points and traces. For example, bus stops, subway stops, important intersections, etc. can be used to extract "important words" in the to-be-completed trajectory and historical trajectory and construct a vector representation. The principle is similar to keeping the nouns, predicates, etc. in the sentence, and removing the function words, articles, etc. that have little to do with the meaning of the sentence, which has little effect on the meaning of the sentence.

[0035] For the trajectory enhancement module, its essential principle comes from text enhancement. In the text enhancement model, through the learning of different types of massive texts, the rules of vocabulary in different contexts can be learned, so that through the trained network model, the missing sentence can be completed. Based on this, in this application, through the big data screening and collection of user historical trajectory information, the common travel modes of users in a certain area can be gradually obtained. These modes can be regarded as the differences in the regional environment, travel time, and travel mode. It is a text corpus of different topics. Based on the training and learning of the collected corpus, the model obtained can effectively learn the context law between different points and traces, thereby realizing the completion of the missing trajectory.

[0036] For the corpus completion module, since the technical theory of trajectory enhancement comes from the field of natural language processing, the enhanced trajectory can only come from "corpus". The coverage of the corpus determines the usability of the system to a certain extent. For example, if all the "corpus" requirements in this system are user action trajectories from A city B district, then it is wrong to enhance a sparse trajectory from A city C district provided by the service side. Therefore, in order to ensure the usability of the system and provide the coverage of the system to all users, the system needs to construct a trajectory corpus completion module to continuously interact with other systems (such as car navigation systems, mobile phone map navigation systems), etc., through continuous The updated and supplemented user's trajectory enables the system to accurately complete different types of sparse trajectories, so that the model learned by the entire system has better generalization capabilities.

[0037] For the trajectory storage and access module, it is used to store the enhanced trajectory obtained through the passage to provide basic data support for services such as travel portraits of people and intelligent transportation. For example, each track in this module can be stored in the format of "track ID, track time, track", where the track ID is used as the only primary key (key) in the key-value pair for retrieval and query, so as to obtain the track time and A collection of related specific points.

[0038] In the embodiments of the present application, the user trajectory is enhanced through the above system architecture. Even if the user does not activate the positioning service, the platform can estimate the user's location, thereby reducing the space range that the user may be currently interested in and improving the recommendation hit rate. By digging out the daily travel trajectory of users, greatly enriching the profile system of specific users, it is possible to realize services such as recommending users' travel routes, and inferring the visiting preferences of users such as shopping, tourism, leisure, etc. The corresponding service recommendation.

[0039] The embodiments of the present application provide an information processing method, device, storage medium, and electronic equipment. Wherein, the information processing device may be specifically integrated in an electronic device or server with computing capability such as a tablet PC (Personal Computer) or a mobile phone that has a storage unit and is installed with a microprocessor.

[0040] Detailed descriptions are given below. It should be noted that the sequence numbers of the following embodiments are not intended to limit the preferred order of the embodiments. In this solution, a self-attention mechanism is introduced, which can mimic the internal process of biological observation behavior, that is, a mechanism that aligns internal experience and external sensation to increase the fineness of observation in some areas. The self-attention mechanism can quickly extract important features of sparse data in trajectory big data, and capture the internal correlation of data or features to analyze and process trajectory big data, so as to achieve the purpose of intelligently reviewing user trajectories. See figure 2 , figure 2 This is a schematic flowchart of the information processing method provided in the embodiments of this application. The specific process of the information processing method can be as follows:

[0041] 101. Obtain the to-be-completed trajectory in the current time period.

[0042] Among them, the current time period may be a time period during which the trajectory of the user's action needs to be effectively complemented, such as one day or one hour. The trajectory to be complemented is a sparse trajectory that needs to be effectively complemented, and may include one or more traces of known position information. In practical applications, the trace can be obtained based on the location data of the LBS of the user terminal device. For example, the location data of the LBS can be data generated when users report their relative locations through mobile devices when they perform applications such as location search and location sharing through various mobile devices.

[0043] In some embodiments, the trajectory internal information fusion can be performed through the self-attention mechanism to be completed. This fusion is based on the recorded position information in the trajectory in the current time period to perform the initial completion of the trajectory, so as to strengthen each time slice. The spatial correlation of the known traces. That is, when acquiring the to-be-completed trajectory in the current time period, the following process may be included:

[0044] Get the known points and traces in the current time period;

[0045] Determine the initial trajectory at least based on the known points in the current time period;

[0046] The internal information of the initial trajectory is fused through the self-attention mechanism to obtain the to-be-completed trajectory.

[0047] In an embodiment, in the trajectory formed by known points in the current time period, some time slices may not have a point trace, so a weighted sum can be performed by using several features of the surrounding points during the initial completion. The obtained new trace is used as a first completion of this point.

[0048] Since time and location are discrete expressions, it is not conducive to deep learning for gradient updates. Therefore, in this embodiment, it is the problem of converting the position information of the dots into the "words" in the sentence in natural language processing, and the time information of the dots is converted into the problem of "position order" in the sentence. Steal each discrete value into a high-dimensional continuous vector in order to directly participate in subsequent network calculations. That is, in some embodiments, when fusing the internal information of the initial trajectory through the self-attention mechanism to obtain the trajectory to be completed, the following process may be included:

[0049] Construct a vector representation of the known points in the initial trajectory;

[0050] According to the vector representation of the known points, determine the similarity between the two known points in the initial trajectory;

[0051] Based on the similarity between the two known points, determine the third attention value of each known point in the initial trajectory with respect to other known points, where the third attention value is used to reflect each The attention degree of a little trace to other known points in its trajectory;

[0052] According to the third attention value, the vector representation of the known points in the initial trajectory is adjusted to obtain the trajectory to be completed.

[0053] Specifically, refer to image 3 , image 3 This is a schematic diagram of the processing flow of the trajectory enhancement module provided in this embodiment. The trajectory enhancement module mainly uses the neural network of the attention mechanism to learn and train the input word vector (it can be a word vector in one-hot format, or a fixed-length word embedding vector generated by the pre-training module) to obtain a Can be used for trajectory enhancement models. The trajectory enhancement module mainly includes three parts: history encoder (History Encoder), self-encoding module (Current Encoder) and decoding module (Current Encoder). The attention mechanism used in the core structure of the three modules. The attention function is designed to map a query and a set of key-value pairs to an output. The output is the weighted sum of these values. The weight is determined by the query using the corresponding key. Calculation. The trajectory embedding module is used to generate a vector representation of the dots in the trajectory, for example, it can convert the features in the space and time dimensions of the dots into a fixed-length vector representation. Among them, the trajectory embedding module can be integrated in the trajectory enhancement module or as a separate module.

[0054] reference image 3 In the current encoder part, when constructing the vector representation of the known points in the initial trajectory, the time representation vector and location representation vector of each trace can be constructed separately, and then the time representation vector and location representation vector of each trace Add and merge to get the vector representation of each trace.

[0055] Since the original time feature vector and location representation vector are indeed not of the same dimension, for example, time can be (hours, minutes), location is latitude and longitude, etc. Therefore, a simple neural network can be used to specify the same number of neurons to become a vector of the same dimension. For example, the time representation vector and location representation vector of each point can be directly represented in the same dimension. For example, the time representation vector is the order position of the point in the track, the first point is "1", and the location is a grid The ID.

[0056] When fusing the internal information of the initial trajectory, the self-attention mechanism is introduced to obtain the features that need to be focused on from the initial trajectory, which is generally referred to as the focus of attention, and then more attention resources are devoted to this feature. In order to obtain more detailed information of the target that needs to be paid attention to, and suppress other useless information, to select the more critical information for the current trajectory from a large number of information, so as to better express each trace and the relationship between the traces. relationship.

[0057] reference Figure 4 In the specific implementation, first construct the embedding vector of the longitude and latitude of the trace (ie, spatial position) and the embedding vector of the location of the trace in the trajectory (ie, time), and then input it to the multi-head attention self-learning module. The relationship between the dots in the trajectory, and then the output vector is aggregated and normalized by multi-head vector. Among them, when performing vector aggregation, the concat method in Python can be used to splice the vectors horizontally or vertically to obtain the spliced ​​vector. When performing vector normalization, the softmax activation function can be used to normalize the output vector to transform the component into a value between [0,1].

[0058] In the attention mechanism, each trace has 3 different vectors, which are Query vector (Q), Key vector (K) and Value vector (V), all of which are 64 in length. They are the embedding vector X of the word multiplied by three different weight matrices W through 3 different weight matrices Q , W K , W V get. Among them, three weight matrix W Q , W K , W V The dimensions are the same, for example, the dimensions can be: 512 x 64.

[0059] In specific implementation, the input traces can be converted into embedding vectors, and then three vectors of Q, K, and V can be obtained according to the embedding vector. Calculate a correlation score (representing similarity) for each trace with other traces, that is, socre = Q * K. In order to stabilize the gradient, the activation function softmax can be used to perform numerical normalization processing on each score. Multiply the normalized value by the Value vector V of each trace to obtain the weighted score V of each input vector, and add the final output result: Z = sum(V), as the Input the attention vector of the dots, and the attention value corresponding to each dot can be obtained by processing the attention vector.

[0060] 102. Determine the first correlation between the trajectory feature of the trajectory to be completed and the trajectory feature of the historical fusion trajectory, where the historical fusion trajectory is determined based on the trajectory information in the historical time period.

[0061] In this embodiment, the historical fusion trajectory has the same period length as the trajectory to be completed. For example, they are all a day, a certain time period of the day (such as 8:00-20:00 on Thursday), etc. The trajectory characteristics of the trajectory to be complemented may include: the spatial identification information of each known point in the trajectory to be complemented, and the position ranking in the trajectory to be complemented. The trajectory characteristics of the historical fusion trajectory include: the spatial identification information of each point in the historical fusion trajectory, and the position order in the historical fusion trajectory. Wherein, the spatial identification information may be geographic location, such as latitude and longitude information. Position sorting can characterize the timing information of the trace in the track.

[0062] When determining the first correlation between the trajectory feature of the trajectory to be completed and the trajectory feature of the historical fusion trajectory, it is specifically possible to construct a vector representation of each point in the trajectory to be completed and the historical fusion trajectory to obtain multiple first features Vector and multiple second feature vectors, and then determine the first similarity between each first feature vector and each second feature vector, and determine the trajectory feature and history fusion of the to-be-completed trajectory based on the first similarity The first correlation between the trajectory features of the trajectory.

[0063] In this embodiment, there may be multiple ways to determine the historical fusion trajectory based on the trajectory information in the historical time period. For example, in one embodiment, when determining the historical fusion trajectory based on the trajectory information in the historical time period, it includes:

[0064] Collect trajectory information in the historical time period;

[0065] According to the specified time period and the track information, construct multiple historical tracks;

[0066] Align multiple historical trajectories according to time, determine the most frequently occurring point trace under the same time slice from the aligned multiple historical trajectories, and construct the target historical trajectory according to the point trace with the highest frequency under the same time slice;

[0067] Through the self-attention mechanism, the internal information of the target historical trajectory is fused to obtain the historical fusion trajectory.

[0068] Specifically, the long-term historical track information can be fused. Among them, the historical period may be the past month, the past week, and so on. For example, if the historical period is one month in the past, the specified time period can be one day.

[0069] Since the historical trajectory is also sparse, the frequent patterns in the history can be mined, that is, the trajectory points with the most occurrences in each time slice of the historical trajectory are first extracted to form a target historical trajectory. For example, refer to image 3 The historical encoder part provides three-day historical trajectories such as trajectories P1, P2, P(m-1), etc. Among them, the trajectory P(m-1) represents the trajectory of any day except trajectories P1 and P2 in the historical period, and m is an integer greater than 2. Track P1 includes known points T5, T9, T13; track P2 includes known points T3, T7, T11; track P(m-1) includes known points T1, T5, T9, T13, T16. First, align the user's daily trajectory according to time, and then extract the most frequently visited locations in each time slice to obtain the target historical trajectory. Since the historical trajectories such as P1, P2, P(m-1) come from different days, the spatial correlation is weakened, so the internal information of the historical trajectory is fused through the self-attention mechanism to select the historical trajectory from the numerous information. Vectors represent more critical information.

[0070] In this embodiment, when the internal information of the target historical trajectory is fused through the self-attention mechanism to obtain the historical fusion trajectory, the vector representation of each trace in the target historical trajectory can be specifically constructed, and then the target can be determined according to the vector representation of each trace. The similarity between two points in the historical trajectory, and based on the similarity between the two points, determine the fourth attention value of each point in the target historical trajectory with respect to the fourth attention value of other points, The fourth attention value is used to reflect the degree of attention of each point in the target historical trajectory to other points in the trajectory, and finally, the vector of the points in the target historical trajectory is determined according to the fourth attention value. Means to adjust to get the historical fusion track.

[0071] Among them, when constructing the vector representation of the points in the historical trajectory of the target, the time representation vector and the location representation vector of each point can be constructed separately, and then the time representation vector and the location representation vector of each trace can be added to obtain the result. The vector representation of each trace.

[0072] reference Figure 4 In the specific implementation, the embedding vector of the longitude and latitude of the dot trace and the embedding vector of the position of the dot trace in the trajectory can be constructed for the historical trajectory, and then input into the multi-head attention self-learning module to learn the relationship between the dots and traces in the trajectory , Perform multi-head vector aggregation and standardization on the output vector for multi-head vector aggregation and standardization. Then, the output result is passed to the next stage through the feedforward neural network, and the enhancement vector after the historical trajectory self-learning is output (that is, the historical fusion trajectory).

[0073] In an embodiment, when determining the historical fusion trajectory based on the trajectory information in the historical time period, it may specifically include:

[0074] Collect trajectory information in the historical time period;

[0075] According to the specified time period and the track information, construct multiple historical tracks;

[0076] Through the attention mechanism, multiple historical trajectories are merged to obtain historical fusion trajectories.

[0077] Specifically, the attention mechanism can be directly used to fuse the internal information of multiple historical trajectories divided by a specified time period to obtain features that need to be focused on in the historical trajectories, and then more attention resources can be devoted to this feature. Obtain more detailed information of the target that needs to be paid attention to and suppress other useless information, so as to select the more critical information for the vector representation of the historical trajectory from the numerous information.

[0078] 103. Fuse the historical fusion trajectory and the to-be-completed trajectory according to the first correlation relationship to obtain the target fusion trajectory.

[0079] In practical applications, due to the sparsity of observations, the credibility of the to-be-completed trajectory obtained based on the known traces is low. Therefore, it can be processed through the attention mechanism to perform the trajectory between the to-be-completed trajectory and the historical fusion trajectory. Information fusion, explicit use of historical trajectory features to extract information that meets the spatial constraints of the current time period. That is, when fusing the historical fusion trajectory and the trajectory to be completed according to the first correlation relationship, candidate points and traces that satisfy the spatial constraints of the trajectory to be completed can be extracted from the historical fusion trajectory to obtain the first candidate point and trace set, and then The trajectory to be complemented is processed according to the first correlation and the first candidate point and trace set, so as to obtain the target fusion trajectory.

[0080] Further, when the trajectory to be complemented is processed according to the first correlation and the first candidate point and trace set, the first target point and trace may be determined from the first candidate point and trace set based on the first correlation, and according to the first The target point trace is adjusted to the point trace information of the trajectory to be complemented to obtain the target fusion trajectory. In an embodiment, when adjusting the track information of the track to be complemented according to the first target track, the following procedures may be specifically included:

[0081] Based on the first similarity, determine the first attention value of each point in the trajectory to be completed with respect to each point in the historical fusion trajectory, where the first attention value is used to reflect the history of each point in the trajectory to be completed The attention degree of each trace in the fusion trajectory;

[0082] Determine the corresponding candidate trace from the first candidate trace set in the descending order of the first attention value as the first target trace;

[0083] According to the position sorting of the first target point trace in the historical fusion trajectory and the spatial identification information corresponding to the first target point, a corresponding point trace is generated at a corresponding position in the trajectory to be completed.

[0084] Among them, the spatial identification information may be geographic location, such as latitude and longitude information; the position sorting may represent the time sequence information of the first target point trace in the historical fusion trajectory.

[0085] Specifically, based on the size of the attention value of each point in the trajectory to be complemented with respect to each point in the historical fusion trajectory, determine the focus of the point and trace of the to-be-completed trajectory from the first candidate point and trace set, and Filter out the characteristics of the point trace in the historical fusion trajectory (that is, time information and spatial position information), and determine the corresponding position in the to-be-completed trajectory to construct a new point trace.

[0086] Continue to refer Figure 4 , Can input the to-be-completed trajectory and the historical fusion trajectory into the multi-head vector decoding and coding attention learning module to learn the relationship between the traces of the trajectories, and perform multi-head vector aggregation and standardization on the output vector. Then, the output result is passed to the next stage through the feedforward neural network, and the to-be-completed trajectory (that is, the target fusion trajectory) after fusing the historical trajectory features is output.

[0087] 104. Determine a second correlation between the trajectory feature of the target fusion trajectory and the trajectory feature of the trajectory to be complemented.

[0088] In this embodiment, the trajectory characteristics of the trajectory to be complemented include: the spatial identification information of each known point in the trajectory to be complemented, and the position ranking in the trajectory to be complemented; the trajectory characteristics of the target fusion trajectory Including: the spatial identification information of each point in the target fusion trajectory, and the position sorting in the target fusion trajectory. Wherein, the spatial identification information may be geographic location, such as latitude and longitude information. Position sorting can characterize the timing information of the trace in the track.

[0089] When determining the second correlation between the trajectory feature of the target fusion trajectory and the trajectory feature of the trajectory to be complemented, the vector representation of each point in the trajectory to be complemented and the target fusion trajectory can be constructed separately to obtain multiple The third feature vector and multiple fourth feature vectors, and then determine the second similarity between each third feature vector and each fourth feature vector, and based on the second similarity, determine the trajectory feature of the target fusion trajectory and The second correlation between the trajectory features of the trajectory to be completed.

[0090] 105. According to the second correlation and the sampling time of the point trace in the to-be-completed trajectory, perform point-and-trace completion processing on the to-be-completed trajectory to obtain a complementary trajectory.

[0091] Specifically, refer to image 3 , The completion processing is to complete the missing points and traces, and the target fusion trajectory output by the encoding module and the original to-be-completed trajectory can be processed by mutual attention mechanism to output the complementary trajectory. In some embodiments, according to the second correlation and the sampling time of the point trace in the to-be-compensated trajectory, the point-and-trace completion processing is performed on the to-be-compensated trajectory to obtain the complementary trajectory, which may specifically include:

[0092] According to the second correlation and the position sorting of the points in the target fusion trajectory, the candidate points are determined from the target fusion trajectory to obtain a second candidate point set;

[0093] According to the sampling time and the second set of candidate points and traces, the point and trace complement processing is performed on the trace to be complemented to obtain the complementary trace.

[0094] Specifically, when the candidate points are determined from the target fusion trajectory according to the second correlation and the positions of the points in the target fusion trajectory, and the second set of candidate points is obtained, the second similarity may be used The second attention value of each point in the trajectory to be complemented with respect to each point in the target fusion trajectory is determined, where the second attention value is used to reflect that each point in the trajectory to be complemented Attention to every little trace. Then, according to the second attention value and the position sorting of the points in the target fusion trajectory, candidate points are determined from the target fusion trajectory to obtain a second candidate point set.

[0095] According to the sampling time and the second set of candidate points and traces, when the trace completion processing is performed on the trajectory to be complemented, it may be based on the known traces in the trajectory to be complemented and the sampling time. The second target point trace is determined in the second candidate point trace set, and then based on the sampling time and the spatial identification information corresponding to the second target point trace, a corresponding point trace is generated at the corresponding position in the to-be-complemented trajectory to be complemented Perform dot and trace completion processing.

[0096] Continue to refer Figure 4 , The to-be-completed trajectory and the target fusion trajectory can be input into the multi-head vector decoding and coding attention learning module to learn the relationship between the traces of the trajectories, and the output vector can be aggregated and standardized by the multi-head vector. Then, the output result is passed to the next stage through the feedforward neural network for vector aggregation and standardization, and the output vector is normalized through the normalization function (such as the softmax activation function), and the normalized component is selected. The dot trace is used as the output, and the enhanced trajectory is obtained (ie, the complement trajectory).

[0097] In this solution, the sampling time of the to-be-completed trajectory point is used as one of the inputs in the attention mechanism, and the to-be-completed trajectory fused with historical information is used as the other input in the attention mechanism, and the midpoint of the trajectory is completed The sampling time and missing traces are processed by attention mechanism to obtain the complement value of missing trace points. Among them, the complement value is a feature component, which represents the index of the missing trace in the entire trace collection. According to the index, the only actual geographic location point can be found, that is, the geographic location of the trace, such as (East longitude 113.935, North latitude 22.542). In the specific calculation, the missing track point can be represented by a special character "unknown", and its feature can be understood as a vector of all zeros.

[0098] For example, the user’s trajectory has a total of four days. The historical trajectory of the first three days is dense and the trajectory of the fourth day is very sparse. It is hoped that through the learning of the historical trajectory of the first three days, he can infer the points that did not appear in the trajectory of the fourth day. position. Then, you can use the trajectory of the fourth day as the target, take a point every 10 minutes, and change this trajectory into a sentence. For example, the trajectory that departs at 9:00 in the morning and arrives at the destination at about 10:00, according to 10 Minute sampling, the track is a 6-word sentence. However, due to the sparse trajectory, some of the words may not be available, and these missing words need to be learned from the historical trajectory. On the other hand, the historical trajectory of the first three days is not necessarily from 9:00 to 10:00 in the morning. Therefore, the historical data can be sampled every 10 minutes according to the time of the day, and then according to the order in the sentence. , And use the trace sampling time on the fourth day as input to complete the trace on the fourth day.

[0099] The information processing method provided by the embodiments of the present application obtains the trajectory to be completed in the current time period, and merges the historical trajectory according to the first correlation between the trajectory feature of the trajectory to be completed and the trajectory feature of the historical fusion trajectory Fuse with the trajectory to be complemented to obtain the target fusion trajectory; according to the second correlation between the trajectory feature of the target fusion trajectory and the trajectory feature of the trajectory to be complemented, and the sampling time of the midpoint of the trajectory to be complemented, Do point and trace completion processing for the entire trajectory to obtain a complementary trajectory. In this solution, by learning the user's trajectory, the missing value of the similar trajectory can be complemented, thereby improving the density of the trace information in the trajectory; in addition, combining the correlation between the trace features in the trajectory, And use this correlation to increase attention to useful information, reduce attention to useless information, and improve the accuracy of track completion information.

[0100] In this application, the trajectory is enhanced to improve the usability of location data. For individuals, it can promote location-based personalized recommendations and the accuracy of advertising; for cities, it can be used for urban population monitoring and crowd flow prediction, and has an important role in alleviating traffic congestion, abnormal events, and environmental pollution. In order to support more mobility-oriented applications, this invention will elaborate on the composition architecture and main technical principles of a trajectory enhancement system, and enhance the availability of mobility data by recovering missing mobility data from collected sparse records.

[0101] To facilitate better implementation of the information processing method provided in the embodiment of the present application, the embodiment of the present application also provides an apparatus based on the foregoing information processing method. The meanings of the nouns are the same as in the above information processing method, and the specific implementation details can refer to the description in the method embodiment.

[0102] See Figure 5 , Figure 5 This is a schematic structural diagram of an information processing device provided by an embodiment of this application, where the processing device may include: an acquisition unit 301, a first determination unit 302, a fusion unit 303, a second determination unit 304, and a processing unit 305. The details can be as follows:

[0103] The acquiring unit 301 is configured to acquire the to-be-completed trajectory in the current time period;

[0104] The first determining unit 302 is configured to determine the first correlation between the trajectory feature of the to-be-completed trajectory and the trajectory feature of the historical fusion trajectory, wherein the historical fusion trajectory is obtained based on the trajectory information in the historical time period;

[0105] The fusion unit 303 is configured to fuse the historical fusion trajectory and the to-be-completed trajectory according to the first correlation relationship to obtain a target fusion trajectory;

[0106] The second determining unit 304 is configured to determine a second correlation between the trajectory feature of the target fusion trajectory and the trajectory feature of the trajectory to be complemented;

[0107] The processing unit 305 is configured to perform point and trace completion processing on the to-be-complemented trajectory according to the second correlation and the sampling time of the point trace in the to-be-compensated trajectory to obtain a complementary trajectory.

[0108] In some embodiments, the fusion unit 303 may be used for:

[0109] Extracting candidate points and traces that satisfy the spatial constraints of the to-be-completed trajectory from the historical fusion trajectory to obtain a first candidate point and trace set;

[0110] The trajectory to be complemented is processed according to the first correlation and the first candidate point trace set to obtain a target fusion trajectory.

[0111] In some embodiments, when processing the to-be-completed trajectory according to the first correlation relationship and the first candidate trace set, the fusion unit 303 may be specifically configured to:

[0112] A first target spot is determined from the first candidate spot set based on the first correlation, and the spot information of the track to be complemented is adjusted according to the first target spot.

[0113] In some embodiments, the trajectory characteristics of the trajectory to be complemented include: the spatial identification information of each known point in the trajectory to be complemented, and the position ranking in the trajectory to be complemented; Trajectory features include: the spatial identification information of each point in the historical fusion trajectory, and the position ranking in the historical fusion trajectory; the first determining unit 302 can be used for:

[0114] Separately constructing a vector representation of each trace in the to-be-completed trajectory and the historical fusion trajectory to obtain multiple first feature vectors and multiple second feature vectors;

[0115] Determine the first similarity between each first feature vector and each second feature vector to obtain the first correlation;

[0116] When the first target point trace is determined from the first candidate point trace set based on the first correlation, and the point trace information of the track to be complemented is adjusted according to the first target point trace, the fusion The unit 303 can be further used for:

[0117] Based on the first degree of similarity, determine the first attention value of each point in the to-be-completed trajectory with respect to each point in the historical fusion trajectory, where the first attention value is used to reflect the to-be-filled The degree of attention of each point in the full trajectory to each point in the historical fusion trajectory;

[0118] Determining a corresponding candidate spot from the first candidate spot set in the descending order of the first attention value as the first target spot;

[0119] According to the position ranking of the first target point trace in the historical fusion trajectory and the spatial identification information corresponding to the first target point, a corresponding point trace is generated at a corresponding position in the trajectory to be completed.

[0120] In some embodiments, the processing unit 305 may be used to:

[0121] Determining candidate points from the target fusion trajectory according to the second correlation and the positional ranking of the points in the target fusion trajectory to obtain a second candidate point and trace set;

[0122] According to the sampling time and the second set of candidate points and traces, performing point and trace completion processing on the to-be-completed trajectory to obtain a complementary trajectory.

[0123] In some embodiments, the trajectory characteristics of the trajectory to be complemented include: the spatial identification information of each known point in the trajectory to be complemented, and the position ranking in the trajectory to be complemented; The trajectory features include: the spatial identification information of each point in the target fusion trajectory, and the position ranking in the target fusion trajectory; the second determining unit 304 can be used for:

[0124] Separately constructing a vector representation of each trace in the trajectory to be completed and the target fusion trajectory to obtain multiple third feature vectors and multiple fourth feature vectors;

[0125] Determine the second degree of similarity between each third feature vector and each fourth feature vector to obtain the second correlation relationship;

[0126] When the candidate points are determined from the target fusion trajectory according to the second correlation and the positions of the points in the target fusion trajectory to obtain a second candidate point and trace set, the processing unit 305 may be specifically configured to:

[0127] Based on the second similarity, the second attention value of each point in the trajectory to be complemented with respect to each point in the target fusion trajectory is determined, where the second attention value is used to reflect the to-be-filled trace The degree of attention of each point in the full trajectory to each point in the target fusion trajectory;

[0128] According to the second attention value and the position ranking of the points in the target fusion trajectory, candidate points are determined from the target fusion trajectory to obtain a second candidate point set.

[0129] In some embodiments, when the trace completion processing is performed on the to-be-completed trajectory according to the sampling time and the second candidate trace set, the processing unit 305 may be further configured to:

[0130] Determining a second target point trace from the second candidate point trace set based on the known point trace in the trajectory to be completed and the sampling time;

[0131] Based on the sampling time and the spatial identification information corresponding to the second target point trace, a corresponding point trace is generated at a corresponding position in the to-be-compensated trajectory, so as to perform point-and-trace completion processing on the to-be-compensated trajectory.

[0132] In some embodiments, the obtaining unit 301 may be used to:

[0133] Get the known points and traces in the current time period;

[0134] Determine the initial trajectory at least based on the known points in the current time period;

[0135] The internal information of the initial trajectory is fused through the self-attention mechanism to obtain the to-be-completed trajectory.

[0136] In some embodiments, when the trajectory internal information fusion is performed on the initial trajectory through the self-attention mechanism to obtain the trajectory to be completed, the acquiring unit 301 may be further configured to:

[0137] Acquiring a vector representation of known points in the initial trajectory;

[0138] Determine the similarity between two known points in the initial trajectory according to the vector representation of the known points;

[0139] Based on the similarity between the two known points, determine the third attention value of each known point in the initial trajectory with respect to other known points, where the third attention value is used to reflect the The attention degree of each point in the initial trajectory to other known points in the trajectory;

[0140] The vector representation of the known points in the initial trajectory is adjusted according to the third attention value to obtain the trajectory to be completed.

[0141] In some embodiments, the information processing apparatus may further include:

[0142] The first collection unit is used to collect trajectory information in a historical time period;

[0143] The first construction unit is used to construct multiple historical trajectories according to the specified time period and the trajectory information;

[0144] The second construction unit is used to align the multiple historical trajectories according to time, determine the most frequently occurring point trace under the same time slice from the aligned multiple historical trajectories, and according to the point trace with the highest frequency under the same time slice Point trace construction to get the target historical trajectory;

[0145] The first trajectory fusion unit is used for fusing the internal information of the target historical trajectory through the self-attention mechanism to obtain the historical fusion trajectory.

[0146] In some embodiments, the first trajectory fusion unit may be specifically used for:

[0147] Construct a vector representation of each point in the target historical trajectory;

[0148] According to the vector representation of each point trace, determine the similarity between two points in the target historical trajectory;

[0149] Based on the similarity between the two points, the fourth attention value of each point in the target historical trajectory with respect to other points is determined, where the fourth attention value is used to reflect each point in the target historical trajectory The attention of the trace to other points in its trajectory;

[0150] The vector representation of the dot trace in the target historical trajectory is adjusted according to the fourth attention value to obtain a historical fusion trajectory.

[0151] In some embodiments, the information processing apparatus may further include:

[0152] The second collecting unit is used to collect track information in the historical time period;

[0153] The third construction unit is used to construct multiple historical tracks according to the specified time period and the track information;

[0154] The second trajectory fusion unit is used for fusing the multiple historical trajectories through the attention mechanism to obtain a fusion trajectory.

[0155] The information processing device provided in this embodiment obtains the trajectory to be completed in the current time period, and combines the historical fusion trajectory with the first correlation between the trajectory feature of the trajectory to be completed and the trajectory feature of the historical fusion trajectory. The trajectory to be complemented is fused to obtain the target fusion trajectory; according to the second correlation between the trajectory feature of the target fusion trajectory and the trajectory feature of the trajectory to be complemented, and the sampling time of the midpoint trace of the trajectory to be complemented, the trajectory to be completed The trajectory is processed by point and trace complementation to obtain the complement trajectory. In this solution, the method realizes the missing value complement of similar trajectories by learning the user trajectory, thereby improving the density of the dot and trace information in the trajectory; in addition, it combines the characteristics of the dot and trace in the trajectory. Relevance, and use this relevance to increase attention to useful information, reduce attention to useless information, and improve the accuracy of track completion information.

[0156] The embodiments of the present application also provide an electronic device, and the electronic device may specifically be a terminal device such as a smart phone or a tablet computer. Such as Image 6 As shown, the electronic device may include a radio frequency (RF, Radio Frequency) circuit 601, a memory 602 including one or more computer-readable storage media, an input unit 603, a display unit 604, a sensor 605, an audio circuit 606, and wireless protection. A true (WiFi, Wireless Fidelity) module 607 includes a processor 608 with one or more processing cores, a power supply 609 and other components. Those skilled in the art can understand, Image 6 The structure of the electronic device shown in does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different component arrangement. among them:

[0157] The RF circuit 601 can be used for receiving and sending signals in the process of sending and receiving information or talking. In particular, after receiving the downlink information of the base station, it is processed by one or more processors 608; in addition, the uplink data is sent to the base station. . Generally, the RF circuit 601 includes, but is not limited to, an antenna, at least one amplifier, a tuner, one or more oscillators, a Subscriber Identity Module (SIM) card, a transceiver, a coupler, and a low noise amplifier (LNA, Low Noise Amplifier), duplexer, etc. In addition, the RF circuit 601 can also communicate with the network and other devices through wireless communication. The wireless communication can use any communication standard or protocol, including but not limited to Global System of Mobile communication (GSM), General Packet Radio Service (GPRS), Code Division Multiple Access (CDMA, Code Division Multiple Access), Wideband Code Division Multiple Access (WCDMA, Wideband Code Division Multiple Access), Long Term Evolution (LTE), Email, Short Messaging Service (SMS, Short Messaging Service), etc.

[0158] The memory 602 may be used to store software programs and modules, and the processor 608 executes various functional applications and data processing by running the software programs and modules stored in the memory 602. The memory 602 may mainly include a program storage area and a data storage area. The program storage area may store an operating system, an application program required by at least one function (such as a sound playback function, an image playback function, etc.), etc.; Data created by the use of electronic devices (such as audio data, phone book, etc.), etc. In addition, the memory 602 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other volatile solid-state storage devices. Correspondingly, the memory 602 may further include a memory controller to provide the processor 608 and the input unit 603 to access the memory 602.

[0159] The input unit 603 can be used to receive inputted number or character information, and generate keyboard, mouse, joystick, optical or trackball signal input related to user settings and function control. Specifically, in a specific embodiment, the input unit 603 may include a touch-sensitive surface and other input devices. A touch-sensitive surface, also called a touch screen or a trackpad, can collect user touch operations on or near it (for example, the user uses any suitable objects or accessories such as fingers, stylus, etc.) on the touch-sensitive surface or on the touch-sensitive surface. Operation near the surface), and drive the corresponding connection device according to the preset program. Optionally, the touch-sensitive surface may include two parts: a touch detection device and a touch controller. Among them, the touch detection device detects the user's touch position, and detects the signal brought by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device, converts it into contact coordinates, and then sends it To the processor 608, and can receive and execute the command sent by the processor 608. In addition, multiple types of resistive, capacitive, infrared, and surface acoustic waves can be used to realize touch-sensitive surfaces. In addition to the touch-sensitive surface, the input unit 603 may also include other input devices. Specifically, other input devices may include, but are not limited to, one or more of a physical keyboard, function keys (such as volume control buttons, switch buttons, etc.), trackball, mouse, and joystick.

[0160] The display unit 604 may be used to display information input by the user or information provided to the user and various graphical user interfaces of the electronic device. These graphical user interfaces may be composed of graphics, text, icons, videos, and any combination thereof. The display unit 604 may include a display panel. Optionally, the display panel may be configured in the form of a liquid crystal display (LCD, Liquid Crystal Display), an organic light emitting diode (OLED, Organic Light-Emitting Diode), etc. Further, the touch-sensitive surface can cover the display panel. When the touch-sensitive surface detects a touch operation on or near it, it is transmitted to the processor 608 to determine the type of the touch event, and then the processor 608 displays the display panel according to the type of the touch event. Corresponding visual output is provided on the panel. Although in Image 6 In the above, the touch-sensitive surface and the display panel are used as two independent components to realize the input and input functions. However, in some embodiments, the touch-sensitive surface and the display panel can be integrated to realize the input and output functions.

[0161] The electronic device may also include at least one sensor 605, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor. The ambient light sensor can adjust the brightness of the display panel according to the brightness of the ambient light, and the proximity sensor can close the display panel and/or when the electronic device is moved to the ear. Backlight. As a kind of motion sensor, the gravity acceleration sensor can detect the magnitude of acceleration in various directions (usually three-axis), and can detect the magnitude and direction of gravity when it is stationary. It can be used to identify mobile phone posture applications (such as horizontal and vertical screen switching, related Games, magnetometer posture calibration), vibration recognition related functions (such as pedometer, percussion), etc.; as for other sensors such as gyroscopes, barometers, hygrometers, thermometers, infrared sensors, etc., which can be configured for electronic devices, we will not Repeat it again.

[0162] The audio circuit 606, speakers, and microphones can provide an audio interface between the user and the electronic device. The audio circuit 606 can transmit the received audio data converted electrical signal to the speaker, and the speaker converts it into a sound signal for output; on the other hand, the microphone converts the collected sound signal into an electrical signal, which is received by the audio circuit 606 and then converted The audio data is processed by the audio data output processor 608, and then sent to, for example, another electronic device through the RF circuit 601, or the audio data is output to the memory 602 for further processing. The audio circuit 606 may also include an earplug jack to provide communication between a peripheral earphone and an electronic device.

[0163] WiFi is a short-range wireless transmission technology. Electronic devices can help users send and receive emails, browse web pages, and access streaming media through WiFi module 607. It provides users with wireless broadband Internet access. although Image 6 The WiFi module 607 is shown, but it is understandable that it is not a necessary component of the electronic device, and can be omitted as needed without changing the essence of the invention.

[0164] The processor 608 is the control center of the electronic device. It uses various interfaces and lines to connect the various parts of the entire mobile phone. By running or executing the software programs and/or modules stored in the memory 602, and calling data stored in the memory 602, Perform various functions of electronic equipment and process data to monitor the mobile phone as a whole. Optionally, the processor 608 may include one or more processing cores; preferably, the processor 608 may integrate an application processor and a modem processor, where the application processor mainly processes the operating system, user interface, and application programs, etc. , The modem processor mainly deals with wireless communication. It can be understood that the foregoing modem processor may not be integrated into the processor 608.

[0165] The electronic device also includes a power source 609 (such as a battery) for supplying power to various components. Preferably, the power source can be logically connected to the processor 608 through a power management system, so that functions such as charging, discharging, and power management are realized through the power management system. The power supply 609 may also include any components such as one or more DC or AC power supplies, recharging systems, power failure detection circuits, power converters or inverters, and power status indicators.

[0166] Although not shown, the electronic device may also include a camera, a Bluetooth module, etc., which will not be repeated here. Specifically, in this embodiment, the processor 608 in the electronic device loads the executable file corresponding to the process of one or more application programs into the memory 602 according to the following instructions, and the processor 608 runs and stores the executable file The application program in the memory 602, thereby realizing various functions:

[0167] Obtain the trajectory to be completed in the current time period; determine the first correlation between the trajectory feature of the trajectory to be completed and the trajectory feature of the historical fusion trajectory, wherein the historical fusion trajectory is obtained based on the trajectory information in the historical time period; The first correlation is to fuse the historical fusion trajectory and the trajectory to be completed to obtain the target fusion trajectory; determine the second correlation between the trajectory feature of the target fusion trajectory and the trajectory feature of the trajectory to be completed; according to the second correlation and The sampling time of the point trace in the trajectory to be complemented is processed by the point trace complement processing on the trajectory to be complemented to obtain the complementary trajectory.

[0168] The electronic device provided by the embodiment of the application realizes the missing value completion of the similar trajectory by learning the user trajectory, thereby improving the density of the dot trace information in the trajectory; in addition, it combines the characteristics of the dot trace in the trajectory And use this correlation to increase attention to useful information, reduce attention to useless information, and improve the accuracy of track completion information.

[0169] The embodiment of the present application also provides a server, and the server may specifically be an application server. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or it can provide cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, and intermediate Cloud servers for basic cloud computing services such as software services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart speaker, a smart watch, etc., but it is not limited to this. The terminal and the server can be directly or indirectly connected through wired or wireless communication, which is not limited in this application. Such as Figure 7 As shown, the server may include a radio frequency (RF, Radio Frequency) circuit 701, a memory 702 including one or more computer-readable storage media, a processor 704 including one or more processing cores, and components such as a power supply 703. . Those skilled in the art can understand, Figure 7 The server structure shown in does not constitute a limitation on the server, and may include more or fewer components than shown in the figure, or a combination of some components, or a different component arrangement. among them:

[0170] The RF circuit 701 can be used for receiving and sending signals in the process of sending and receiving information or talking. In particular, after receiving the downlink information of the base station, it is processed by one or more processors 704; in addition, the uplink data is sent to the base station. . Generally, the RF circuit 701 includes, but is not limited to, an antenna, at least one amplifier, a tuner, one or more oscillators, a Subscriber Identity Module (SIM) card, a transceiver, a coupler, and a low noise amplifier (LNA, Low Noise Amplifier), duplexer, etc. In addition, the RF circuit 701 can also communicate with the network and other devices through wireless communication. The wireless communication can use any communication standard or protocol, including but not limited to Global System of Mobile communication (GSM), General Packet Radio Service (GPRS), Code Division Multiple Access (CDMA, Code Division Multiple Access), Wideband Code Division Multiple Access (WCDMA, Wideband Code Division Multiple Access), Long Term Evolution (LTE), Email, Short Messaging Service (SMS, Short Messaging Service), etc.

[0171] The memory 702 can be used to store software programs and modules, and the processor 704 executes various functional applications and data processing by running the software programs and modules stored in the memory 702. The memory 702 may mainly include a program storage area and a data storage area. The program storage area may store an operating system, an application program required by at least one function (such as a sound playback function, an image playback function, etc.), etc.; Data (such as audio data, phone book, etc.) created by the use of the server. In addition, the memory 702 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other volatile solid-state storage devices. Correspondingly, the memory 702 may further include a memory controller to provide the processor 704 and the input unit 703 to access the memory 702.

[0172] The processor 704 is the control center of the server. It uses various interfaces and lines to connect various parts of the entire mobile phone. It executes by running or executing software programs and/or modules stored in the memory 702, and calling data stored in the memory 702. Various functions of the server and processing data, so as to monitor the mobile phone as a whole. Optionally, the processor 704 may include one or more processing cores; preferably, the processor 704 may integrate an application processor and a modem processor, where the application processor mainly processes the operating system, user interface, and application programs, etc. , The modem processor mainly deals with wireless communication. It can be understood that the foregoing modem processor may not be integrated into the processor 704.

[0173] The server also includes a power source 703 (such as a battery) for supplying power to various components. Preferably, the power source may be logically connected to the processor 704 through a power management system, so that functions such as charging, discharging, and power management are realized through the power management system. The power supply 703 may also include any components such as one or more DC or AC power supplies, a recharging system, a power failure detection circuit, a power converter or inverter, and a power status indicator.

[0174] Specifically, in this embodiment, the processor 704 in the server loads the executable file corresponding to the process of one or more application programs into the memory 702 according to the following instructions, and the processor 704 runs and stores the executable file in the memory. The application in 702 can realize various functions:

[0175] Obtain the trajectory to be completed in the current time period; determine the first correlation between the trajectory feature of the trajectory to be completed and the trajectory feature of the historical fusion trajectory, where the historical fusion trajectory is obtained based on the trajectory information in the historical time period; The first correlation is to fuse the historical fusion trajectory and the trajectory to be completed to obtain the target fusion trajectory; determine the second correlation between the trajectory feature of the target fusion trajectory and the trajectory feature of the trajectory to be completed; according to the second correlation and The sampling time of the point trace in the to-be-complemented trajectory is processed by the point-and-trace completion processing on the to-be-complemented trajectory to obtain the complementary trajectory.

[0176] The server provided by the embodiment of the application realizes the missing value completion of the similar trajectory by learning the user trajectory, thereby improving the density of the dot and trace information in the trajectory; in addition, it combines the characteristics of the dot and trace in the trajectory. And use this correlation to increase the attention to useful information, reduce attention to useless information, and improve the accuracy of track completion information.

[0177] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above-mentioned embodiments can be completed by instructions, or by instructions to control related hardware. The instructions can be stored in a computer-readable storage medium. And loaded and executed by the processor.

[0178] To this end, an embodiment of the present application provides a computer-readable storage medium in which a plurality of instructions are stored, and the instructions can be loaded by a processor to execute the steps in any information processing method provided in the embodiments of the present application. . For example, the instruction can perform the following steps:

[0179] A first set of entities associated with the first text and a second set of entities associated with the second text are determined based on a preset knowledge base, the preset knowledge base including a knowledge representation composed of entities, relationships between entities and entity attributes Determine the entity correlation between the first group of entities and the second group of entities according to the knowledge representation; according to the association relationship between each word in the first text, each of the second text The association relationship between the words and the association relationship between the words in the first text and the words in the second text, determine the attention of each word in the first text and the second text with respect to other words Strength value, wherein the attention value is used to reflect the attention degree of each word in the first text and the second text to other words; at least according to the attention value and the entity relevance, determine The text relevance of the first text and the second text.

[0180] For the specific implementation of the above operations, please refer to the previous embodiments, which will not be repeated here.

[0181] Wherein, the storage medium may include: read only memory (ROM, Read Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk, etc.

[0182] Since the instructions stored in the storage medium can execute the steps in any information processing method provided in the embodiments of this application, it can achieve what can be achieved by any information processing method provided in the embodiments of this application. For the beneficial effects, refer to the previous embodiment for details, which will not be repeated here.

[0183] The information processing method, device, storage medium, and electronic equipment provided by the embodiments of the application are described in detail above. Specific examples are used in this article to illustrate the principles and implementations of the application. The description of the above embodiments is only It is used to help understand the method and core idea of ​​this application; at the same time, for those skilled in the art, according to the idea of ​​this application, there will be changes in the specific implementation and scope of application. In summary, this specification The content should not be construed as a limitation on this application.