Method and apparatus for predicting sorting of data in time sequence

Abstract

Provided are a method and apparatus for predicting the sorting of data in time sequence, the method specifically comprising: a first model acquiring a data set and determining a first predicted time sequence pair according to the data set; a second model determining an associated time sequence pair according to the first predicted time sequence pair and a logic rule; iterating the first model according to the associated time sequence pair until the first predicted time sequence pair is the same as the associated time sequence pair outputted by the second model, and the first model outputting a second predicted time sequence pair; and according to the second predicted time sequence pair, sorting the time sensitivity of generated data. By "combining a machine learning model with a logic rule", the technical problem that "for time-sensitivity sorting models, it is difficult to validate whether a ranking conforms to a time sequence in the real world" is solved, and the accuracy of a machine learning model and a logic rule can be effectively improved compared with separate use.

Description

A method and device for predicting the order of time series data Technical Field

[0001] The present invention relates to the field of data recognition, and in particular to a method and device for sorting and predicting time series data. Background Art

[0002] Over time, attribute values ​​in data may become outdated and inaccurate. Even worse, only partially reliable timestamps may be available. Generally, when we say a timestamp is reliable, we mean that it is accurate and correct, and that it indicates that the attribute value at that time was correct and up-to-date.

[0003] Existing technology requires machine learning.

[0004] However, machine learning has its limitations. For time-sensitive ranking models, it is difficult to prove whether the rankings are consistent with the chronological order in the real world.

[0005] Summary of the Invention

[0006] In view of the above problems, the present application is proposed to provide a method and apparatus for predicting the ranking of time series data that overcomes the above problems or at least partially solves the above problems, including:

[0007] A method for predicting the ranking of time-series data is provided, wherein the method is applied to predict the timeliness of data through an artificial intelligence model; wherein the artificial intelligence model includes a first model and a second model, including:

[0008] The first model acquires a data set and determines a first predicted time sequence pair based on the data set;

[0009] The second model determines an associated time sequence pair based on the first predicted time sequence pair and a logic rule;

[0010] Iterating the first model according to the associated time sequence pair until the first predicted time sequence pair is the same as the associated time sequence pair output by the second model, and the first model outputs a second predicted time sequence pair;

[0011] The timeliness of the generated data is sorted according to the second prediction time order.

[0012] Preferably, the step of the first model acquiring a data set and determining a first predicted time sequence pair based on the data set comprises:

[0013] The first model acquires a data set; wherein the data set includes data and attribute values ​​corresponding thereto; the data set includes at least two groups of data;

[0014] The first model determines a first predicted time sequence pair according to the attribute value of the data.

[0015] Preferably, the step of determining the first predicted time sequence pair according to the attribute value of the data by the first model includes:

[0016] The first model obtains the confidence of the initial time sequence pair according to the attribute value of the data;

[0017] The first model filters the initial time sequence according to the confidence of the initial time sequence pair to obtain the first predicted time sequence pair.

[0018] Preferably, the step of determining the associated time sequence pairs according to the first predicted time sequence pairs and the logic rule by the second model includes:

[0019] The second model executes a pursuit procedure according to the first prediction time sequence and logic rules to obtain effective pursuit and invalid pursuit;

[0020] The second model obtains the associated time sequence pairs according to the effective pursuit.

[0021] Preferably, the second model executes the pursuit procedure according to the first prediction time sequence and logic rules to obtain the steps of effective pursuit and invalid pursuit, including:

[0022] The second model chasing procedure derives the effective chasing and the invalid chasing by using logic rules and verification data in the first prediction time sequence.

[0023] Preferably, the step of iterating the first model according to the association time sequence comprises:

[0024] The first model is iterated using the associated time-sequential pairs and the data set as input.

[0025] Preferably, the step of sorting the timeliness of generated data according to the second prediction time order includes:

[0026] A total order of the attribute values ​​of the data is generated according to the second predicted time order.

[0027] To implement this application, a device for predicting the order of time series data is also provided, comprising:

[0028] A first prediction time sequence pair module, configured to acquire a data set using the first model and determine a first prediction time sequence pair based on the data set;

[0029] an associated time sequence pair module, configured for the second model to determine an associated time sequence pair based on the first predicted time sequence pair and a logic rule;

[0030] a second prediction time sequence pair module, configured to iterate the first model according to the associated time sequence pair until the first prediction time sequence pair is the same as the associated time sequence pair output by the second model, and the first model outputs a second prediction time sequence pair;

[0031] A timeliness sorting module is used to sort the timeliness of the generated data according to the second prediction time sequence.

[0032] To implement the present application, a computer electronic device is also included, including a processor, a memory, and a computer program stored in the memory and capable of running on the processor. When the computer program is executed by the processor, the steps of the method for sorting and predicting temporal data are implemented.

[0033] To implement the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for sorting and predicting time series data are implemented.

[0034] This application has the following advantages:

[0035] In the embodiments of the present application, compared to the prior art problem that "it is difficult for time-based ranking models to prove whether the ranking conforms to the time sequence in the real world", the present application provides a solution of "combining machine learning models and logical rules", specifically: the first model acquires a data set and determines a first predicted time sequence pair based on the data set; the second model determines an associated time sequence pair based on the first predicted time sequence pair and logical rules; the first model is iterated based on the associated time sequence pairs until the first predicted time sequence pair is the same as the associated time sequence pair output by the second model, and the first model outputs a second predicted time sequence pair; the time-based ranking of the generated data is generated based on the second predicted time sequence pair. The present application solves the problem of "it is difficult for time-based ranking models to prove whether the ranking conforms to the time sequence in the real world" by "combining machine learning models and logical rules". Compared with when used separately, the framework of the present application can improve the accuracy of the machine learning model and logical rules. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solution of the present application, the following is a brief introduction to the drawings required for the description of the present application. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0037] FIG1 is a flowchart of a method for predicting the order of time series data provided by an embodiment of the present application;

[0038] FIG2 is a structural framework diagram of a method for sorting and predicting time series data provided by an embodiment of the present application;

[0039] FIG3 is a data set diagram of a method for sorting and predicting time series data provided by an embodiment of the present application;

[0040] FIG4 is a structural block diagram of a device for predicting the order of time series data provided by an embodiment of the present application;

[0041] FIG5 is a schematic structural diagram of a computer device provided by an embodiment of the present invention;

[0042] 12. Computer device; 14. External device; 16. Processing unit; 18. Bus; 20. Network adapter; 22. I / O interface; 24. Display; 28. Memory; 30. Random access memory; 32. Cache memory; 34. Storage system; 40. Program / utility; 42. Program module. DETAILED DESCRIPTION

[0043] To make the objectives, features, and advantages of this application more readily apparent, the present application is further described below in conjunction with the accompanying drawings and specific embodiments. It is apparent that the embodiments described are only a portion of the embodiments of this application, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments in this application without inventive effort are also within the scope of protection of this application.

[0044] By analyzing existing technologies, the inventors discovered that over time, data attribute values ​​may become outdated and inaccurate. Worse still, only partially reliable timestamps may be available. Generally, when we say a timestamp is reliable, we mean that it is accurate and correct, and that it indicates that the attribute value at that time was correct and up-to-date. Besides mechanical reasons (malicious attacks or hardware failures), the following logical reasons are the main timestamp-related issues in data quality and explain the lack of reliable timestamps.

[0045] 1. Lack of timestamps. Timestamps may not be recorded at all. For example, in one electronic health database, only 16 of 26 relational tables were timestamped. Even if a relational table has a timestamp, it may be incomplete.

[0046] 2. Inaccurate timestamps. Timestamps may be too coarse, resulting in unreliable time sorting. A spreadsheet may be submitted to the OA system by an employee multiple times in a day. If the forms are recorded with date stamps, it is unclear which form (all on the same day) is the latest. Similar problems are often encountered in hospital data, where only dates are recorded. Another example involves inconsistent timestamp strength (for example, minutes vs. days). If two values ​​have timestamps of "12-8-2021" and "12-8-202120:41" respectively, it is unclear which one is the latest.

[0047] 3. Incorrect timestamps. Many factors can cause incorrect timestamps. For example, in medical data, an X-ray machine has many asynchronous modules, each with a local clock and a local buffer. Because values ​​are queued in the buffer before being recorded, there may be a discrepancy between when the value is actually updated and when it is recorded.

[0048] Therefore, the problem to be solved by this application is that in the absence of a complete timestamp, this application needs to determine the timeliness of the data and the latest level of the data.

[0049] As shown in Figure 3, a picture of a dataset is shown. Consider a set of records belonging to the same entity. Their attribute values ​​may be outdated and inaccurate. Even worse, only partially reliable timestamps may be available. In this case, how can we determine the newness of the data attributes? That is, given two records t1 and t2 belonging to the same entity, we need to determine whether the value of attribute A in t1 is newer than the value of attribute A in t2, expressed as t2 < A t1.

[0050] As shown in Figure 3, customer records t1-t6 have been confirmed to refer to the same person, Mary. Each record records some attribute values ​​of her marital status, job, number of children, and SZ (shoe size). The attributes of some of these records are outdated. For example, her job, address, and last name have changed 4 times, 5 times, and 2 times respectively. And only some attribute values ​​may have reliable timestamps. For example, the timestamps of t5[job] and t6[job] may be 2016 and 2019 respectively, indicating that at that time, these attribute values ​​were the latest. However, in the absence of complete timestamps, it is difficult to know whether t2 < LN t6, that is, the last name value of t2 is negated and the last name value of t6 is new. In addition, questions such as what is Mary's latest job need to be sorted.

[0051] Existing technologies address this problem by training a temporal ranking model that ranks data attribute values ​​based on the relevance or importance of the objects. State-of-the-art systems employ deep learning or reinforcement learning and have been used in search engines and machine translation. Ranking models can learn temporal order and, for all records t1t2 and attribute A, determine whether t2 < A t1.

[0052] Another approach is to use logical rules to help us infer time sequences. For example, as shown in Figure 3, shoe sizes for a person typically increase monotonically (before the age of 20), and a person's address may be related to their marital status. These constraints can be abstracted into logical rules to infer time sequences.

[0053] However, both machine learning and logical rules have their limitations.

[0054] For temporal ranking models, it is difficult to prove whether the ranking conforms to the real-world temporal order. For data-driven decision making, we need to ensure that the learned order is reliable. In addition, these methods cannot explain the ordering of objects that follow complex relationships (such as addresses).

[0055] For logical rules, it's difficult to find enough rules to infer the relative order of every pair of values. When testing existing rule-based methods on a dataset with 5% initial timestamps, even the best method could only infer 16.3% of the temporal order, leaving the remaining 78.7% undetermined. Furthermore, it's difficult to generalize rules to handle lexically different but semantically similar values ​​(e.g., marital status: married vs. married).

[0056] A natural question is whether it is possible to combine machine learning models and logical rules in a unified framework, so that we can learn the temporal order and use rules to verify the ranking and improve the model's learning. Compared to using them separately, the framework of this application effectively improves the accuracy of machine learning models and logical rules.

[0057] Therefore, we propose a creator-critic framework for determining temporal order, as shown in Figure 2, and develop a system called GATE to implement it. A unique feature of GATE is its combination of deep learning and logical deduction. Below, we begin with GATE's architecture and then present its overall workflow.

[0058] The ultimate goal of GATE is to obtain a total order for each attribute value. As shown in Figure 2, GATE first discovers logical rules on the data offline to perform logical deduction. It then takes a time series database as input and learns and infers more time-order pairs. More specifically, the learning and deduction process in GATE iteratively executes two stages: the creator and the critic, as shown below.

[0059] It should be noted that, in any embodiment of the present invention, the method is applied to timeliness sorting of data through artificial intelligence model prediction; wherein, the artificial intelligence model includes a first model and a second model.

[0060] 1 , a flowchart of a method for predicting the ranking of time series data according to an embodiment of the present application is shown, which specifically includes the following steps:

[0061] S110, the first model acquires a data set and determines a first predicted time sequence pair based on the data set;

[0062] S120: The second model determines an associated time sequence pair based on the first predicted time sequence pair and a logic rule;

[0063] S130, iterating the first model according to the associated time sequence pair until the first predicted time sequence pair is the same as the associated time sequence pair output by the second model, and the first model outputting a second predicted time sequence pair;

[0064] S140. Sort the timeliness of the generated data according to the second prediction time sequence.

[0065] In the embodiments of the present application, compared to the problem in the prior art that "it is difficult for time-based ranking models to prove whether the ranking conforms to the time sequence in the real world", the present application provides a solution of "combining machine learning models and logical rules", specifically: the first model obtains a data set and determines a first predicted time sequence pair based on the data set; the second model determines an associated time sequence pair based on the first predicted time sequence pair and logical rules; the first model is iterated based on the associated time sequence pair until the first predicted time sequence pair is the same as the associated time sequence pair output by the second model, and the first model outputs a second predicted time sequence pair; the time-based ranking of the generated data is generated based on the second predicted time sequence pair. The present application solves the problem of "it is difficult for time-based ranking models to prove whether the ranking conforms to the time sequence in the real world" by "combining machine learning models and logical rules", and achieves that the framework of the present application can improve the accuracy of the machine learning model and logical rules compared to when used separately.

[0066] Next, the method for predicting the order of time series data in this exemplary embodiment will be further described.

[0067] As described in step S110 above, the first model acquires a data set and determines a first predicted time sequence pair based on the data set.

[0068] In one embodiment of the present invention, the specific process of "the first model acquiring a data set and determining a first prediction time sequence pair based on the data set" in step S110 can be further explained in combination with the following description.

[0069] As described in the following steps, the first model obtains a data set; wherein, the data set contains data and attribute values ​​corresponding thereto; the data set includes at least two groups of data; the first model determines a first predicted time sequence pair based on the attribute values ​​of the data.

[0070] In one embodiment of the present invention, the specific process of the step of "the first model determines the first predicted time sequence pair according to the attribute value of the data" can be further explained in combination with the following description.

[0071] As described in the following steps, the first model obtains the confidence of the initial time sequence pair based on the attribute value of the data; the first model filters the initial time sequence based on the confidence of the initial time sequence pair to obtain the first predicted time sequence pair.

[0072] As an example, the first model screens the initial time sequence pairs according to the confidence of the initial time sequence pairs through a predefined threshold, selects the initial time sequence pairs that are higher than the predefined threshold, and obtains the first predicted time sequence pairs.

[0073] In a specific embodiment, the first model is the creator. At this stage, GATE can use any existing ranking model. Given a record t1t2 and attribute A, the model calculates t2 < A The confidence level of t1 is expressed as conf(t2< A t1) indicates that. When conf(t2< A When t1) is greater than a predefined threshold δ, we say t2< A t1 is a candidate time sequence pair.

[0074] As described in step S120 above, the second model determines associated time sequence pairs based on the first predicted time sequence pairs and logic rules.

[0075] In one embodiment of the present invention, the specific process of "the second model determines the associated time sequence pair based on the first predicted time sequence pair and the logical rule" in step S120 can be further explained in combination with the following description.

[0076] As described in the following steps, the second model executes the pursuit procedure according to the first predicted time sequence and logic rules to obtain effective pursuit and invalid pursuit; the second model obtains the associated time sequence pair according to the effective pursuit.

[0077] As described in the following steps, the second model chasing procedure obtains the valid chasing and invalid chasing by deriving and repairing the first prediction time sequence using logic rules and verification data.

[0078] In a specific embodiment, the second model tracks the first predicted time through logical rules to obtain associated time sequence pairs; wherein the logical rules are deduced from data associated with the first predicted time sequence pairs to obtain associated time sequence pairs.

[0079] In one specific embodiment, the second model is a critic. During this stage, GATE's critic executes a chasing procedure using the logical rules obtained offline to prove and infer more time sequence pairs. Based on the effectiveness of the chasing procedure results, we construct an enhanced training data set, which may include verified time sequence pairs and conflicting order pairs captured by the rules. This information is fed back to the creator, requiring the creator to resolve potential conflicts by modifying their model accordingly. This information is then used for the next round of model learning, allowing the creator to learn from more unknown data and iterate to achieve higher accuracy. With more iterations, the time sequence pairs will gradually expand and eventually become stable and unchanged.

[0080] In a specific embodiment, however ideal, chasing may be expensive if we enumerate all rules in an exhaustive manner. Below we provide an efficient algorithm to implement the chasing procedure.

[0081] A brute-force approach to chasing is to enumerate all rules. If applicable, a chase step is performed until the chase sequence terminates. However, this approach is expensive because enumeration is exponential in nature. To address this challenge, we develop an efficient algorithm for chasing; the key idea is to lazily recall only the rules relevant to the affected fix during the chase.

[0082] To allow for lazy invocation, rules are generated and applied only when invoked by some newly derived temporal pair, rather than being constructed entirely at the start of the chase. Specifically, when a new temporal pair is derived, we check to invoke a rule that satisfies the following conditions: (a) the rule is affected by the inferred fix, (b) the rule has never been invoked before, and (c) the temporal pair inferred by the rule has never been derived before. We maintain a specific data structure to efficiently check these three conditions.

[0083] As an example, the chase program is a program that, given a time series database, infers repairs by using rules and validation data in the data. It uses a set ≤ Σ To track the affected fixes in the chase.

[0084] Specifically, the form of the i-th pursuit step is:

[0085] in, is a logic rule, and h is the application of the logic rule. The application of the logic rule requires the following conditions: (1) all verification data are satisfied and (2) the repair of step i The derived t2< A t1 expands into Similarly, ≤Σ is also expanded accordingly.

[0086] From a set of initial repairs Initially, a chase sequence is:

[0087] A chase sequence terminates in one of the following cases: (1) no more logical rules can be applied. If so, we say it is a valid chase; or (2) a conflict is found in the repair, in which case we say it is an invalid chase.

[0088] Intuitively, when the pursuit ends with a valid result, it can help us infer more temporal order pairs; moreover, if there is no invalid pursuit, it will prove and explain the learned order. When its result is invalid, it detects the conflicting temporal order pairs predicted by the creator.

[0089] It should be noted that repair means that we extend the time sequence pairs by applying rules in the data to derive “repairs”, and these repairs are modeled as temporary time sequence pairs t2< A The set of t1, represented by, where each t2 < A t1 means t2< A t1 is derived by the rules. It's important to note that here we only apply the logical rules satisfied by the "verification data" to derive the "fix." Intuitively, the fix is ​​the logical consequence of the rules and the verification data; that is, as long as the rules and the verification data are correct, the fix is ​​also correct.

[0090] We usually say that a set of repairs is conflict-free if there is no attribute A and record t1t2 such that t2 < A t1 and t1< A t2 exists at the same time.

[0091] Validation data is used to prove the correctness of the repair. We maintain and use a set of validation data. Initially, the validation data can be specified by the user in a small amount. It is then iteratively expanded in the process by the time sequence pairs predicted by the creator above the confidence threshold, or by chasing inference among critics.

[0092] As an example, the second model obtains the associated time sequence pairs based on the effective pursuits, removes invalid pursuits, and verifies the effective pursuits.

[0093] As described in step S130 above, the first model is iterated according to the associated time sequence pair until the first predicted time sequence pair is the same as the associated time sequence pair output by the second model, and the first model outputs the second predicted time sequence pair.

[0094] In one embodiment of the present invention, the specific process of step S130 of "iterating the first model according to the associated time sequence pair until the first predicted time sequence pair is the same as the associated time sequence pair output by the second model, and the first model outputs the second predicted time sequence pair" can be further explained in combination with the following description.

[0095] As described in the following steps, the first model is iterated using the associated time sequence pairs and the data set as input.

[0096] As an example, the associated time sequence pair is used as the input of the first model, and the loop is continuously repeated until the first predicted time sequence pair is the same as the associated time sequence pair output by the second model, that is, the second model no longer outputs content, and the associated time sequence pair after multiple loops is placed in the first model to obtain the second predicted time sequence pair.

[0097] In one embodiment, the associated time-sequence pairs are fed back to the creator, who is then asked to modify their model accordingly to resolve potential conflicts. This information is then used in the next round of model learning, allowing the creator to learn from more unknown data and iterate to achieve higher accuracy. With more iterations, the time-sequence pairs gradually expand and eventually become stable and unchanging.

[0098] As described in step S140 above, the timeliness of the generated data is sorted according to the second prediction time sequence.

[0099] In one embodiment of the present invention, the specific process of "sorting the timeliness of generated data according to the second prediction time sequence" in step S140 can be further explained in combination with the following description.

[0100] As described in the following steps, a total order of the attribute values ​​of the data is generated according to the second predicted time order.

[0101] In a specific embodiment, we verify the effectiveness of the present invention through experiments on real datasets. We implement a system called GATE based on the creator-critic framework and compare the effects of using creators alone and critics alone.

[0102] Experimental results show that GATE converges quickly, requiring fewer than ten iterations on average. For example, on a dataset covering the careers of football players from FIFA-15 to FIFA-22, GATE's F1, MRR (Mean Reciprocal Rank), and MAP (Mean Average Precision) metrics increased from 0.767 to 0.866, 0.786 to 0.857, and 0.752 to 0.809, respectively, demonstrating its ability to infer up-to-date values ​​and produce well-timed rankings. This is because the creator iteratively accumulates training data from the critic, enabling the model to be better trained over more rounds. Furthermore, as the creator's predictions improve, the critic infers more temporal order pairs for the creator in subsequent rounds as enhanced training data. Furthermore, GATE achieves precision and recall of 0.859 and 0.873, respectively, demonstrating that GATE strikes a good balance between the two and is quite accurate.

[0103] GATE outperforms creators and critics on the F1 metric by an average of 7.8% and 34.4%, respectively, reaching peak performance of 11.0% and 45.6%, respectively, significantly improving both approaches. This demonstrates that under our framework, creators and critics mutually benefit. On COM, an open-source dataset of self-employed entrepreneurs in Shenzhen, the critic generated an average of 6,000 new training data points per round, improving GATE's F1 metric from 0.701 to 0.748 after five rounds.

[0104] As an example, we propose GATE, a creator-critic framework that determines temporal order by combining deep learning and logical deduction. GATE iteratively calls on creators to rank the temporal order of attribute values. Critics then verify the creators' rankings and infer more temporal order pairs based on the discovered rules. Critics also generate augmented training data for the creators to improve the accuracy of their rankings in the next round. This process continues, with creators and critics mutually reinforcing each other, until the accuracy of the temporal order cannot be further improved.

[0105] Compared to machine learning-based time-based ranking models, GATE demonstrates that the ranking is reliable and consistent with real-world chronological order through a rule-based tracking process for reviewers. Furthermore, creators can generate more chronological pairs for reviewers to use for logical reasoning.

[0106] Compared to rule-based methods, GATE has stronger generalization and robustness. By logically reasoning about time sequence pairs, it can in turn provide enhanced training data to creators for model optimization.

[0107] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0108] 4 , a device for predicting the order of time series data provided by an embodiment of the present application is shown, which specifically includes the following modules:

[0109] First prediction time sequence pair module 410: used for the first model to acquire a data set and determine a first prediction time sequence pair based on the data set;

[0110] An associated time sequence pair module 420 is used for the second model to determine an associated time sequence pair based on the first predicted time sequence pair and a logic rule;

[0111] A second prediction time sequence pair module 430 is configured to iterate the first model according to the associated time sequence pair until the first prediction time sequence pair is the same as the associated time sequence pair output by the second model, and the first model outputs a second prediction time sequence pair;

[0112] Timeliness sorting module 440: used to sort the timeliness of generated data according to the second prediction time sequence.

[0113] In one embodiment of the present invention, the first prediction time sequence pair module 410 includes:

[0114] Acquisition submodule: used for acquiring a data set using the first model; wherein the data set includes data and attribute values ​​corresponding thereto; the data set includes at least two groups of data;

[0115] The first predicted time sequence pair submodule is used for the first model to determine the first predicted time sequence pair according to the attribute value of the data.

[0116] In one embodiment of the present invention, the first prediction time sequence pair submodule includes:

[0117] Confidence submodule: used for the first model to obtain the confidence of the initial time sequence pair according to the attribute value of the data;

[0118] Screening submodule: used for the first model to screen the initial time sequence according to the confidence of the initial time sequence pair to obtain the first predicted time sequence pair.

[0119] In one embodiment of the present invention, the time sequence association module 420 includes:

[0120] Chasing submodule: used for the second model to execute the chasing procedure according to the first prediction time sequence and logic rules to obtain effective chasing and invalid chasing;

[0121] Effective pursuit submodule: used for the second model to obtain the associated time sequence pair according to the effective pursuit.

[0122] In one embodiment of the present invention, the second prediction time sequence pair module 430 includes:

[0123] Iteration submodule: used for iterating the first model by taking the associated time sequence pairs and the data set as input.

[0124] In one embodiment of the present invention, the timeliness sorting module 440 includes:

[0125] Total order submodule: used for generating a total order of attribute values ​​of the data according to the second prediction time order.

[0126] It should be noted that for the sake of simplicity, the method embodiments are described as a series of actions. However, those skilled in the art should be aware that the embodiments of the present invention are not limited by the order of the actions described, because according to the embodiments of the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present invention.

[0127] There are some overlapping operation steps in this specific embodiment and the above specific embodiment, so this specific embodiment is only briefly described. For other solutions, please refer to the description of the above specific embodiment.

[0128] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0129] 5 , a computer device for a method for predicting the ranking of time series data according to the present invention is shown, which may specifically include the following:

[0130] The computer device 12 is a general-purpose computing device. The components of the computer device 12 may include but are not limited to: one or more processors or processing units 16, memory 28, and a bus 18 connecting different system components (including memory 28 and processing unit 16).

[0131] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures. Examples of these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, an Audio Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.

[0132] The computer device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the computer device 12, including volatile and non-volatile media, removable and non-removable media.

[0133] The memory 28 may include computer system readable media in the form of volatile memory, such as random access memory 30 and / or cache memory 32. The computer device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be used to read and write to a non-removable, non-volatile magnetic medium (commonly referred to as a "hard drive"). Although not shown in FIG. 5 , a magnetic disk drive for reading and writing to a removable non-volatile magnetic disk (such as a "floppy disk"), as well as an optical disk drive for reading and writing to a removable non-volatile optical disk (such as a CD-ROM, DVD-ROM or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data media interfaces. The memory may include at least one program product having a set (e.g., at least one) of program modules 42 configured to perform the functions of the various embodiments of the present application.

[0134] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in a memory. Such program modules 42 include, but are not limited to, an operating system, one or more application programs, other program modules 42, and program data, each of which, or some combination thereof, may include an implementation of a network environment. The program modules 42 generally perform the functions and / or methods of the embodiments described herein.

[0135] The computer device 12 may also communicate with one or more external devices 14 (e.g., a keyboard, pointing device, display 24, camera, etc.), one or more devices that enable an operator to interact with the computer device 12, and / or any device that enables the computer device 12 to communicate with one or more other computing devices (e.g., a network card, modem, etc.). Such communication may occur via an I / O interface 22. Furthermore, the computer device 12 may also communicate with one or more networks (e.g., a local area network (LAN)), a wide area network (WAN), and / or a public network (e.g., the Internet) via a network adapter 20. As shown in FIG5 , the network adapter 20 communicates with the other modules of the computer device 12 via a bus 18. It should be understood that, although not shown in FIG5 , other hardware and / or software modules may be used in conjunction with the computer device 12, including, but not limited to, microcode, device drivers, redundant processing units 16, external disk drive arrays, RAID systems, tape drives, and data backup storage systems 34.

[0136] The processing unit 16 executes various functional applications and data processing by running the programs stored in the memory 28, such as implementing a sorting prediction method for time series data provided in an embodiment of the present application.

[0137] That is, when the above-mentioned processing unit 16 executes the above-mentioned program, it is achieved: the first model obtains a data set and determines a first prediction time sequence pair based on the data set; the second model determines an associated time sequence pair based on the first prediction time sequence pair and logical rules; the first model is iterated according to the associated time sequence pair until the first prediction time sequence pair is the same as the associated time sequence pair output by the second model, and the first model outputs a second prediction time sequence pair; and the timeliness of the generated data is sorted according to the second prediction time sequence pair.

[0138] In an embodiment of the present application, the present application also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, a method for sorting and predicting time series data as provided in all embodiments of the present application is implemented.

[0139] That is, when the program is executed by the processor, it is implemented as follows: the first model obtains a data set and determines a first prediction time sequence pair based on the data set; the second model determines an associated time sequence pair based on the first prediction time sequence pair and logical rules; the first model is iterated based on the associated time sequence pair until the first prediction time sequence pair is the same as the associated time sequence pair output by the second model, and the first model outputs a second prediction time sequence pair; and the timeliness of the generated data is sorted based on the second prediction time sequence pair.

[0140] Any combination of one or more computer-readable media may be employed. A computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0141] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take a variety of forms, including, but not limited to, electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0142] The computer program code for performing the operations of the present application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" language or similar programming languages. The program code can be executed entirely on the operator's computer, partially on the operator's computer, as a separate software package, partially on the operator's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer can be connected to the operator's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet). The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same and similar parts between the various embodiments can be referenced to each other.

[0143] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic inventive concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0144] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or terminal device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or terminal device that includes the element.

[0145] The above is a detailed introduction to a method and device for sorting and predicting time series data provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core idea of ​​the present application. At the same time, for general technical personnel in this field, based on the ideas of the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for sorting and predicting time series data, the method is applied to sort the data according to timeliness through artificial intelligence model prediction; wherein: The artificial intelligence model includes a first model and a second model, and is characterized in that it includes: The first model acquires a data set and determines a first predicted time sequence pair based on the data set; The second model determines the associated time sequence pairs according to the first predicted time sequence pairs and the logic rule; Iterating the first model according to the associated time sequence pair until the first predicted time sequence pair is the same as the associated time sequence pair output by the second model, and the first model outputs a second predicted time sequence pair; The timeliness of the generated data is sorted according to the second prediction time order.

2. The method for ranking and predicting time series data according to claim 1, characterized in that: The step of acquiring a data set by the first model and determining a first predicted time sequence pair according to the data set comprises: The first model acquires a data set; wherein the data set includes data and attribute values ​​corresponding thereto one by one; the data set includes at least two groups of data; The first model determines a first predicted time sequence pair according to the attribute value of the data.

3. The method for predicting the order of time series data according to claim 2, characterized in that: The step of determining the first predicted time sequence pair according to the attribute value of the data by the first model comprises: The first model obtains the confidence of the initial time sequence pair according to the attribute value of the data; The first model screens the initial time sequence according to the confidence of the initial time sequence pair to obtain the first predicted time sequence pair.

4. The method for predicting the ranking of time series data according to claim 1, characterized in that: The step of determining, by the second model, the associated time sequence pairs according to the first predicted time sequence pairs and the logic rule comprises: The second model performs a chasing procedure according to the first prediction time sequence and logic rules to obtain to effective pursuit and ineffective pursuit; The second model obtains the associated time sequence pairs according to the effective pursuit.

5. The method for ranking and predicting time series data according to claim 4, characterized in that: The second model executes the chasing procedure according to the first prediction time sequence and logic rules to obtain the steps of effective chasing and invalid chasing, including: The second model chasing procedure obtains the valid chasing and invalid chasing by deriving and repairing the first prediction time sequence using logic rules and verification data.

6. The method for predicting the ranking of time series data according to claim 1, characterized in that: The step of iterating the first model according to the association time sequence comprises: The first model is iterated using the associated time-ordered pairs and the data set as input.

7. The method for predicting the ranking of time series data according to claim 1, characterized in that: The step of sorting the timeliness of generated data according to the second prediction time sequence includes: A total order of attribute values ​​of the data is generated according to the second predicted time order.

8. A device for predicting the order of time series data, wherein: include: A first prediction time sequence pair module, used for the first model to obtain a data set and determine a first prediction time sequence pair based on the data set; An associated time sequence pair module, used for the second model to determine an associated time sequence pair according to the first predicted time sequence pair and a logic rule; A second prediction time sequence pair module, used for iterating the first model according to the associated time sequence pair, until the first prediction time sequence pair is the same as the associated time sequence pair output by the second model, and the first model outputs a second prediction time sequence pair; A timeliness sorting module is used to sort the timeliness of generated data according to the second prediction time sequence.

9. A computer electronic device, characterized in that: It comprises a processor, a memory and a computer program stored in the memory and capable of running on the processor, wherein when the computer program is executed by the processor, the steps of the method for sorting and predicting time series data as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for sorting and predicting time series data as described in any one of claims 1 to 7 are implemented.

Images