Method and apparatus for constructing human-like answering engine

Abstract

The present invention relates to the technical field of artificial intelligence, and provides a method and apparatus for constructing a human-like answering engine. The method comprises: acquiring a state transition graph of each answering algorithm; respectively merging and integrating input nodes, output nodes, identical key state nodes, and identical state transition paths in all state transition graphs to form an integrated state transition graph; and performing function configuration on the integrated state transition graph to construct a static graph of an answering technique as the human-like answering engine. The method and apparatus for constructing the human-like answering engine provided by the present invention simulate the thought processes of human, and progressively analyze and solve algebraic questions by means of a series of state transitions, so that the answering engine is more flexible and intelligent, and can handle complex and variable question types, thereby expanding the range of answerable questions.

Description

A method and apparatus for constructing a humanoid solution engine [Technical Field]

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a method and apparatus for constructing a humanoid solution engine. [Background Technology]

[0002] Machine problem-solving has long been a vibrant and cutting-edge research topic in the field of artificial intelligence. In recent years, driven by technological advancements in related fields and the combined demand for intelligent education, machine-based algebraic problem-solving has reached a new research peak. Algebraic problems involve calculations based on algebraic relationships, including physics algebra problems, chemistry algebra problems, arithmetic word problems, mathematical algebra problems, algebra problems involving graphs, plane geometry calculation problems, plane geometry algebra calculation problems, function-based algebra problems, and so on. After more than 60 years of development, machine problem-solving has yielded abundant research results. In terms of machine problem-solving theory, state transition theory has been developed. This theory can both explain existing problem-solving algorithms and provide guidance for designing new algorithms. Regarding machine problem-solving algorithms, more than 100 algorithms for solving algebraic problems have been developed. To introduce their main technologies, state transition theory can be used to classify these algorithms into five main methods of representing solution states: methods based on double-frame models, equation-based methods, expression-based methods, Seq2X-based methods, and relation group-based methods. The following sections will briefly describe the theories and algorithms developed for machine-based solutions to algebraic problems.

[0003] 1. State transition theory for machine solving algebraic problems

[0004] The state transition theory of machine learning for solving algebraic problems models the solution process as a cyclical representation of transitions from one state to another using transition techniques. Here, "state" is an abstract representation of the intermediate result states of the algorithm, and "transition techniques" are the methods used to move from one state to another. At the algorithm implementation level, the state is an abstract data structure used to uniformly store a set of intermediate results, while the transition techniques are a set of executable programs with standard interfaces. Based on state transition theory, machine learning algorithms for solving algebraic problems can be represented as "state transition diagrams." These diagrams are graphical representations of the internal structure of existing machine learning algorithms, facilitating not only the analysis of existing methods and algorithms but also the analysis of their future evolutionary direction.

[0005] 2. Existing solution algorithms can be mainly categorized into five types of solution methods.

[0006] (1) Solution method based on the dual-frame model. The core idea of ​​the solution method based on the dual-frame model is to use a knowledge representation framework to store the result of understanding the problem, and to use a strategy framework as the driving force for knowledge reasoning to act on the knowledge representation framework, thereby generating the problem understanding process and the solution result. According to the state transition theory, the state transition diagram of the solution method based on the dual-frame model generally uses two types of frameworks to construct the solution model representing arithmetic word problems. These two types of frameworks are the knowledge representation framework and the reasoning strategy framework. Therefore, the solution method based on the dual-frame model has two main state nodes, namely the state node of the knowledge representation framework and the state node of the reasoning strategy framework. Combining the problem input state node and the solution output state node, the solution method based on the dual-frame model has three main transition techniques: the first transition technique is to use semantic template matching to fill the knowledge contained in the problem text into the knowledge representation framework; the second transition technique is to use problem type matching to select the reasoning strategy framework of the problem; and the third transition technique is to use the reasoning strategy framework to reason and solve the problem to obtain the solution.

[0007] (2) Equation-based solution method. The core idea of ​​the equation-based solution method is to use equations as the equivalent representation of the problem, and to form a system of equations based on these equations. Solving the system of equations is the driving force for finding the answer to the problem. Since solving the system of equations can rely on existing mature mathematical tools, the main stage of solving problems using the equation-centered solution method lies in problem understanding, that is, how to obtain the equations that are equivalent to the problem and form a system of equations. The equation system used by the equation-centered method is a new equivalent representation of the problem proposed after the double-frame model. Compared with the double-frame model method, the equivalent state of the problem in the equation-centered method is simplified, and the process of obtaining the system of equations reduces the reliance on natural language understanding of the problem text. According to the state transition theory, there are five states in the state transition diagram of the equation-based solution method, of which there are three main states: feature text state node, equation template state node, and equation system state node. Therefore, there are four transfer techniques in this type of solution algorithm: the first transfer technique is to obtain the feature text of the question using feature extraction methods; the second transfer technique is to calculate and match the feature text to the equation template; the third transfer technique is to use parameter mapping to transform the equation template into a system of equations; and the fourth transfer technique is to solve the system of equations to obtain the answer to the question.

[0008] (3) Expression-based solution method. The expression-based solution method transforms the problem text into an answer expression composed of numbers and operators for solution. According to state transition theory, the state transition diagram of the expression-based solution method has three main states: quantity operands, expression tree, and expression. Therefore, this state transition diagram has four transition techniques: the first transition technique is to classify and identify the quantities in the problem text to obtain the quantity operands for parameter operations; the second transition technique is to use rules to reason about the quantity operands to obtain the expression tree corresponding to the problem text; the third transition technique is to transform the expression tree into an expression; and the fourth transition technique is to solve the expression to obtain the answer to the problem.

[0009] (4) Seq2X-based solution method. The Seq2X (Sequence-to-X, where X can be Seq, Tree, and DAG, corresponding to sequence, tree, and directed acyclic graph, respectively)-based solution method utilizes a trained Seq2X model to transform the input state of the problem into an answer expression for computation and solution. The underlying assumption of the Seq2X solution method is that a problem can be equivalently represented by an answer expression. This method draws on the "encode-decode" paradigm in the field of natural language processing, and its process mainly involves calculating vectors. According to the state transition theory, the state transition diagram of the Seq2X-based solution algorithm has five state nodes. In addition to the input problem text state node and the output problem answer state node, there are three key state nodes: the text vector sequence state node, the expression vector sequence state node, and the expression state node. Therefore, this type of solution algorithm uses four transition techniques to connect these five state nodes: the first transition technique encodes the given problem text into a sequence of text vectors; the second transition technique transforms the sequence of problem text vectors into a latent vector of the expression; the third transition technique transforms the latent vector of the expression into the answer expression; and the fourth transition technique calculates and solves the answer expression to obtain the solution to the problem. The state transition diagram of the Seq2X-based solution method shows that although subsequent research has developed and integrated many new deep learning techniques and methods, the main focus has been on introducing new techniques and methods for the actions in the state transition diagram to construct new solution algorithms.

[0010] (5) Relationship-based solution method. The relationship-based solution method transforms the input state of the problem into a relationship group, which in turn transforms into a system of equations for solving. The underlying assumption of this method is that an arithmetic word problem can be equivalently represented as a system of equations for solving. The core of the relationship-based solution method is to use relationships as the granularity for understanding the problem, rather than isolated numbers. The main process of understanding the problem is to obtain the relationship group needed for the solution. The algorithm of the relationship-based solution method has a common state, namely the relationship group. Therefore, the state transition diagram of the relationship-based solution method has three main states: text annotation sequence, relationship group, and equation group. The output state of this method is the problem solution. The state transition diagram of the relationship-based solution method has four key transition techniques: the first is to perform part-of-speech tagging on the problem text to obtain a text annotation sequence; the second is to extract relationships from the annotation sequence of the problem text to obtain a relationship group; the third is to transform the relationship group into a system of equations; and the fourth is to solve the equation group to obtain the problem solution. The relationship-based solution method subsequently introduces vector computation, resulting in vectorized representation of the state and transition techniques.

[0011] The above review of the theories and algorithms developed for machine-based algebra problem solving shows that significant progress has been made in this field. However, the results have not yet met the requirements for intelligent education applications, and there is still considerable room for development. Although over 100 algorithms have been proposed in the field of machine-based algebra problem solving, a common practice is that a single paper often aims to use a single algorithm to solve different types of problems. This approach has been proven infeasible, as the range of problems that such algorithms can solve is very limited. [Summary of the Invention]

[0012] This invention provides a method and apparatus for constructing a human-like problem-solving engine to address the limitation of the range of problems that can be solved in the prior art.

[0013] In a first aspect, the present invention provides a method for constructing a human-like problem-solving engine, comprising: obtaining a state transition diagram for each problem-solving algorithm; the state transition diagram includes input nodes, key state nodes, output nodes, and state transition paths between nodes, wherein input nodes correspond to the input states of the problem-solving algorithm, key state nodes correspond one-to-one with the key states of the problem-solving algorithm, output nodes correspond to the output states of the problem-solving algorithm, and state transition paths correspond one-to-one with the transition relationships between states; merging and integrating the input nodes, output nodes, identical key state nodes, and identical state transition paths in all state transition diagrams to form an integrated state transition diagram; and performing the integration... The state transition graph is configured to construct a static graph of the solution technology as a human-like solution engine. The method for configuring the integrated state transition graph includes: configuring input nodes in the integrated state transition graph to receive algebraic problems, analyze their content, and determine the next state transition path; configuring key state nodes in the integrated state transition graph to determine the next state transition path based on the state content generated by the current node; and configuring output nodes in the integrated state transition graph to output the solution to the algebraic problems based on the state content received from the state transition nodes.

[0014] According to a method for constructing a humanoid solution engine provided by the present invention, for input nodes and key state nodes, determining the next state transition path for state transition includes: when a node has multiple state transition paths, determining at least one target state transition path from the multiple state transition paths for state transition based on the currently received state content; when a node has only one state transition path, using the only state transition path for state transition.

[0015] According to a method for constructing a humanoid solution engine provided by the present invention, when the key state node receives state content transferred by multiple state transition paths, data integration processing is performed on all state content.

[0016] According to a method for constructing a humanoid solution engine provided by the present invention, when a node has multiple state transition paths, the node is referred to as a divergent transition node. Based on the currently received state content, at least one target state transition path is determined from the multiple transition paths for state transition. This includes: inputting the currently received state content into a pre-constructed first decision model of the current divergent transition node, and outputting predicted path labels for state transition. The step of constructing the first decision model of any divergent transition node includes: obtaining a decision dataset of any divergent transition node regarding state transition paths; the decision dataset includes the state content received by the any divergent transition node and the corresponding path labels; wherein the path labels include the next state transition path and the node's label; based on the decision dataset, analyzing the rule-driven mapping relationship between the state content and the path labels, and constructing the first decision model of the any divergent transition node.

[0017] According to a method for constructing a human-like solution engine provided by the present invention, when a node has multiple state transition paths, the node is referred to as a divergent transition node; based on the currently received state content, at least one target state transition path is determined from the multiple transition paths for state transition, including: inputting the currently received state content into a pre-trained second decision model of the current divergent transition node, and outputting the predicted path label for state transition; wherein, the step of constructing a second decision model for any divergent transition node includes: obtaining a decision dataset for any divergent transition node regarding state transition paths; the decision dataset includes the state content received by the any divergent transition node and the corresponding path label; wherein, the path label includes the next state transition path and the node's label; training a preset machine learning model using the decision dataset, and using the trained preset machine learning model as the second decision model.

[0018] The method for constructing a humanoid solution engine according to the present invention further includes: acquiring necessary knowledge from a pre-set knowledge base when necessary knowledge is lacking during state transition.

[0019] The method for constructing a humanoid solution engine according to the present invention further includes: updating the configuration of the humanoid solution engine when a new solution algorithm appears.

[0020] According to a method for constructing a humanoid solution engine provided by the present invention, the data integration processing of all state content includes one or more of the following processing methods: information merging, content cleaning, and priority sorting.

[0021] Secondly, the present invention also provides an apparatus for constructing a human-like solution engine, comprising:

[0022] The state transition diagram acquisition module is used to acquire the state transition diagram of each solution algorithm. The state transition diagram includes input nodes, key state nodes, output nodes, and state transition paths between nodes. Input nodes correspond to the input states of the solution algorithm, key state nodes correspond one-to-one with the key states of the solution algorithm, output nodes correspond to the output states of the solution algorithm, and state transition paths correspond one-to-one with the transition relationships between states.

[0023] The integrated state transition diagram construction module is used to merge and integrate the input nodes, output nodes, and identical key state nodes and identical state transition paths in all state transition diagrams to form an integrated state transition diagram.

[0024] A humanoid solution engine construction module is used to configure the integrated state transition diagram to construct a static diagram of the solution technology as a humanoid solution engine; wherein, configuring the integrated state transition diagram includes:

[0025] Configure the input nodes in the integrated state transition diagram to receive algebraic problems, parse their content, and determine the next state transition path for state transition; configure the key state nodes in the integrated state transition diagram to determine the next state transition path based on the state content generated by the current node through state transition; configure the output nodes in the integrated state transition diagram to output the solution to the algebraic problems based on the state content received from the state transition nodes.

[0026] Thirdly, the present invention provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the steps of the method for constructing a humanoid solution engine as described above.

[0027] Fourthly, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for constructing a humanoid solution engine as described above.

[0028] The method and apparatus for constructing a human-like problem-solving engine provided by this invention integrates the state transition diagrams of multiple algorithms to construct an integrated state transition diagram, and then performs functional configuration on the integrated state transition diagram to generate a static diagram of the problem-solving technology as a human-like problem-solving engine. This can further improve the performance and efficiency of problem-solving. Furthermore, this invention imitates the human thought process, using a series of state transitions to gradually analyze and solve algebraic problems, making the problem-solving engine more flexible and intelligent, capable of handling complex and varied problem types, and expanding the range of solvable problems. [Attached Image Description]

[0029] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0030] Figure 1 is a flowchart illustrating the method for constructing a humanoid solution engine provided in an embodiment of the present invention;

[0031] Figure 2 is a schematic diagram of the static diagram of the solution technique constructed using two algorithm state transition diagrams provided in an embodiment of the present invention;

[0032] Figure 3 is a static diagram of the solution technology provided in the embodiment of the present invention;

[0033] Figure 4 is a schematic diagram of the process of solving algebraic problems using a humanoid solution engine provided in an embodiment of the present invention.

[0034] Figure 5 is a schematic diagram of the structure of the device for constructing a humanoid solution engine provided in an embodiment of the present invention;

[0035] Figure 6 is a schematic diagram of the structure of the electronic device provided in an embodiment of the present invention.

Detailed Implementation Methods

[0036] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0037] It should be noted that, in the description of the embodiments of the present invention, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.

[0038] The terms "first," "second," etc., used in this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class, without limiting the number of objects; for example, a first object can be one or more. Furthermore, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects have an "or" relationship.

[0039] The method and apparatus for constructing a humanoid solution engine provided by the embodiments of the present invention are described below with reference to Figures 1-6.

[0040] Figure 1 is a flowchart illustrating the method for constructing a humanoid solution engine provided in an embodiment of the present invention. As shown in Figure 1, the method includes, but is not limited to, the following steps:

[0041] Step 101: Obtain the state transition diagram for each solution algorithm.

[0042] In this invention, the state transition diagram can be pre-constructed. Based on the input state, key state, output state, and transition relationships between states constructed using transition techniques, a state transition diagram for each solution algorithm is built to represent the solution process of each algorithm. The state transition diagram includes input nodes, key state nodes, output nodes, and state transition paths between nodes. Input nodes correspond to the input states of the solution algorithm, key state nodes correspond one-to-one with the key states of the solution algorithm, output nodes correspond to the output states of the solution algorithm, and state transition paths correspond one-to-one with the transition relationships between states.

[0043] Specifically, referring to the state transition theory described in the background section above, this invention models the solution process for algebraic problems as a cyclical representation of a state transitioning from one state to another using transition techniques. Here, "state" is an abstract representation of the intermediate result states of the algebraic problem-solving algorithm, and "transition technique" is the method for transitioning from one state to another. At the algorithm implementation level, the state is an abstract data structure used to uniformly store a set of intermediate results, while the transition technique is a set of executable programs with standard interfaces. Based on the state transition theory, the algorithm for solving algebraic problems by machine can be represented as a "state transition diagram," which is a graphical representation of the internal structure of existing machine-solved algorithms.

[0044] The state transition diagram in this invention includes input nodes, key state nodes, output nodes, and state transition paths between nodes.

[0045] The input nodes correspond to the input states, indicating where the algorithm begins, and typically receive input data (e.g., in algebraic problems); the key state nodes correspond one-to-one with the key states, representing important milestones in the solution process; the output nodes correspond to the output states, indicating where the algorithm ends, and outputting the final solution result; the state transition paths correspond one-to-one with the transition relationships, which are edges connecting different nodes, representing the transition process from one state to another.

[0046] In the state transition diagram, new solution methods and solutions can be defined: the solution method describes the machine solution at an abstract level, which only considers the connection path from the problem to the understanding state, without considering the technical methods on the connecting edges between nodes; the solution algorithm considers the specific path from the starting state to the ending state, including the technical methods on the connecting edges between nodes.

[0047] Referring to the background information above, over 100 algorithms for solving algebraic problems have been developed. To illustrate their key technologies, the state transition theory of machine learning can be used to categorize these algorithms into five main methods for representing solution states: double-frame model-based methods, equation-based methods, expression-based methods, Seq2X-based methods, and relation group-based methods. These will not be elaborated upon further here.

[0048] This invention analyzes existing solution algorithms based on state transition theory, and constructs a static diagram of solution technology, which will be further explained below.

[0049] Step 102: Merge and integrate the input nodes, output nodes, identical key state nodes, and identical state transition paths in all state transition diagrams to form an integrated state transition diagram.

[0050] By merging the state transition graphs of all individual algorithms into an integrated state transition graph, redundancy is reduced and efficiency is improved by merging duplicate key state nodes and state transition paths.

[0051] It is understandable that different key state nodes and state transition paths should be retained so that the integrated state transition graph, which is constructed by fusing all state transition graphs, retains the algorithmic paths of all state transition graphs.

[0052] Furthermore, it should be noted that the integrated state transition diagram formed by the present invention may also include other new solution paths, thereby increasing the solution range of algebraic problems.

[0053] Step 103: Configure the integrated state transition diagram to construct a static diagram of the solution technology as a humanoid solution engine; wherein, the method for configuring the integrated state transition diagram includes: configuring input nodes in the integrated state transition diagram to receive algebraic problems and perform content parsing on the algebraic problems to determine the next state transition path for state transition; configuring key state nodes in the integrated state transition diagram to determine the next state transition path for state transition based on the state content generated through state transition received by the current node; configuring output nodes in the integrated state transition diagram to output the solution content of the algebraic problems based on the state content input by the received state transition nodes.

[0054] It should be noted that the next state transition path may include multiple state transition paths.

[0055] The process of configuring the integrated state transition diagram is essentially designing and optimizing the algorithmic logic for machines to solve algebraic problems. Specific steps include:

[0056] (1) Configure input node

[0057] The input node is the entry point for the entire solution process, responsible for receiving algebraic problems and performing initial content analysis. This step mainly involves:

[0058] Input format parsing: Identify and understand the structure and format of the input algebraic problems, such as equations, inequalities, expressions, graphs, and their components such as variables, constants, and operators.

[0059] State initialization: Based on the parsed problem content, an initial state is created. This state contains all the necessary information about the problem and is used in the subsequent state transition process.

[0060] (2) Configure critical state nodes

[0061] Key state nodes represent important decision points in the algorithm, with each node corresponding to a specific state. At these nodes, the algorithm determines its next action based on the current state. The configuration process includes:

[0062] Status assessment: Analyze the information of the current status to determine whether a certain condition is met or a certain goal is achieved.

[0063] Transition path selection: Based on the state evaluation results, a suitable transition path is selected, that is, deciding which state node to transition to next. This may involve complex algorithmic logic; for example, different mathematical rules or algorithms may need to be applied for specific types of algebraic problems.

[0064] (3) Configure output node

[0065] The output node is the endpoint of the algorithm, used to output the final solution based on the received state content. The configuration process mainly includes:

[0066] Result generation: Based on the current state, generate the final solution. This may include operations such as calculation, simplification, and formatting.

[0067] Output formatting: Convert results into a user-friendly format, for example, convert mathematical expressions into an easy-to-read form.

[0068] (4) Other configuration strategies

[0069] Throughout the configuration process, the focus is on ensuring that the algorithm can correctly parse the problem, effectively perform state transitions, and generate accurate solutions. Additionally, configuration optimization may include:

[0070] Error handling: Add an exception handling mechanism to ensure that the algorithm can provide reasonable feedback when it encounters problems that cannot be parsed or error states.

[0071] Extensible Design: When a new solution algorithm emerges, the present invention can configure and update the humanoid solution engine according to the new solution algorithm, so that the algorithm can flexibly adapt to algebraic problems of different types and difficulties, and expand the functionality of the algorithm by adding new state nodes and transition paths.

[0072] With the above configuration, the static graph solution technology can become a powerful and flexible tool for automatically solving a wide range of algebraic problems. Furthermore, this invention mimics the human thought process, using a series of state transitions to gradually analyze and solve algebraic problems, making the solution engine more flexible and intelligent, capable of handling complex and varied problem types, and expanding the range of solvable problems.

[0073] Based on the above embodiments, as an optional embodiment, the method for constructing a humanoid solution engine provided by the present invention, for input nodes and key state nodes, determines the next state transition path for state transition, including: when a node has multiple state transition paths, determining at least one target state transition path from the multiple state transition paths for state transition based on the currently received state content; when a node has only one state transition path, using the only state transition path for state transition. When input nodes and key state nodes have multiple state transition paths, the present invention can refer to input nodes and key state nodes as divergent transition nodes. It is understood that divergent transition nodes in the present invention generally refer to input nodes or key state nodes with multiple state transition paths.

[0074] The core of building a human-like problem-solving engine lies in mimicking the human thought process to achieve a step-by-step understanding and solution to a problem. This method particularly emphasizes the ability to dynamically select transition paths based on the received state content at the divergent exit node, thereby achieving a more intelligent and flexible problem-solving strategy. Specifically, this method includes the following two key scenarios.

[0075] Multi-path selection scenario: When the outgoing node has multiple possible state transition paths, the engine needs to be able to intelligently select at least one of the most suitable target state transition paths based on the currently received state content. This means that the engine must not only understand the current state but also predict the possible outcomes of different paths, thereby making the optimal choice. This ability is similar to the process by which humans make decisions based on experience and knowledge when facing complex problems.

[0076] Single-path deterministic scenario: Conversely, when there is only a single state transition path for the diverging outgoing node, the engine does not need to perform complex path selection, but directly performs state transitions along that path. This usually occurs when the solution path to the problem is relatively clear or when there is only one reasonable operation at a certain stage.

[0077] Whether it's multi-path selection or single-path determination, both are based on a deep understanding of the state content. By progressively analyzing the problem, the engine can navigate the state space and find the optimal path from the problem to the solution. This approach not only improves problem-solving efficiency but also enhances the engine's adaptability and robustness, enabling it to handle mathematical problems of various complexities and types, and even problems from other domains.

[0078] Optionally, the method for constructing a humanoid solution engine provided by the present invention further includes: when the key state node receives state content transferred from multiple state transition paths, performing data integration processing on all state content; such a key state node can be called a convergence node. It should be noted that a key state node may be both a convergence node and a divergence node.

[0079] In building a human-like problem-solving engine, besides the selection and determination of state transition paths mentioned above, effectively integrating and processing the state content received by key state nodes from multiple state transition paths is also a crucial step. This process aims to fuse information from different paths to form a more comprehensive and deeper understanding, supporting subsequent decision-making and problem-solving.

[0080] By integrating state information from different paths, this invention allows the engine to acquire more multi-dimensional information, thereby making more accurate and comprehensive judgments. This helps avoid erroneous decisions caused by incomplete information.

[0081] Furthermore, the state information along different paths often reflects different aspects or perspectives of the problem. Integrating this information can help the engine understand the essence of the problem more deeply, improving the accuracy and depth of the solution.

[0082] Optionally, the data integration processing method of the present invention for all state content includes one or more of the following processing methods: information merging, content cleaning, and priority sorting. Further explanation follows.

[0083] Information merging refers to the process of integrating state information collected from multiple state transition paths into a unified whole. This process typically includes data fusion, which involves merging similar or related information to reduce duplication and extract the most essential content.

[0084] Content cleaning aims to improve data quality by performing preliminary cleaning and preprocessing on the collected state content to remove irrelevant or redundant information and ensure data quality and consistency.

[0085] Prioritization is used to determine which status information is more important or urgent for solving the current problem, and to sort the status information according to priority.

[0086] Based on the above embodiments, as an optional embodiment, the method for constructing a humanoid solution engine provided by the present invention, for a divergent transition node, determines at least one target state transition path from multiple transition paths to perform state transition based on the currently received state content, including: inputting the currently received state content into a pre-constructed first decision model of the current divergent transition node, and outputting the predicted path label to perform state transition.

[0087] The steps of constructing a first decision model for any divergent outgoing node include: obtaining a decision dataset for any divergent outgoing node regarding the state transition path; the decision dataset includes the state content received by the any divergent outgoing node and the corresponding path label; wherein the path label includes the label of the next state transition path and the node; based on the decision dataset, analyzing the rule-driven mapping relationship between the state content and the path label, and constructing a first decision model for the any divergent outgoing node.

[0088] Specifically, when determining that the data type of the input state content is a data type with a clear structure and identifiable patterns (e.g., keyword text, syntactic and semantic text), the rule-driven first decision model has a greater advantage.

[0089] Taking keyword text as an example, keywords or phrases closely related to path selection are identified from the decision dataset. The frequency and location of keyword occurrences and their correlations are analyzed, and keyword combinations or patterns corresponding to specific path labels are summarized.

[0090] Taking syntactic and semantic text as an example, we analyze the syntactic structure in state content, identify basic components such as subject-verb-object and part-of-speech change patterns, and understand the basic framework of sentences; we label the semantic roles of each component in the sentence and understand their functions and meanings in the sentence; based on semantic roles and syntactic structure, we identify semantic patterns that match specific path tags.

[0091] Based on the above analysis, a series of decision rules are defined. These rules explicitly indicate under what conditions the input should meet to select which path label for state transition (i.e., rule-driven mapping). For example, "If the input contains keywords A and B, and the syntactic structure conforms to pattern X, then choose path Y."

[0092] When the first decision model receives new state content, it first performs keyword and syntactic-semantic analysis to extract key information; then it matches the extracted information with predefined decision rules to find if there are any matching rules; if a matching rule is found, it outputs the corresponding path label according to the rule instruction to guide the state transition.

[0093] Based on the above embodiments, as an optional embodiment, the method for constructing a humanoid solution engine provided by the present invention, for a divergent transition node, determines at least one target state transition path from multiple transition paths to perform state transition based on the currently received state content, including: inputting the currently received state content into a pre-trained second decision model of the current divergent transition node, and outputting the predicted path label to perform state transition.

[0094] The step of constructing a second decision model for any divergent outgoing node includes: obtaining a decision dataset for any divergent outgoing node regarding the state transition path; the decision dataset includes the state content received by the any divergent outgoing node and the corresponding path label; wherein the path label includes the next state transition path and the label of the node; training a preset machine learning model using the decision dataset, and using the trained preset machine learning model as the second decision model.

[0095] Understandably, the machine learning model is the core of the entire mechanism. The model can be any type of supervised learning, such as Support Vector Machine (SVM), Random Forest, or Deep Neural Network (DNN), depending on the characteristics of the data and the complexity of the problem. When the data type of the input state data is unsuitable for the first decision model, a second decision model based on machine learning is used to determine path labels for path transition.

[0096] Based on the above embodiments, as an optional embodiment, the method for constructing a humanoid solution engine provided by the present invention further includes: obtaining necessary knowledge from a pre-set knowledge base when necessary knowledge is lacking during the state transition process.

[0097] The knowledge base is a crucial component of a human-like solution engine. It stores a vast amount of organized and categorized domain knowledge, factual information, concept definitions, formulas, and principles, covering a wide range of fields from basic science to advanced theories. The construction and maintenance of the knowledge base is an ongoing process, requiring continuous updates and expansion to adapt to the emergence of new knowledge and changes in problem scenarios.

[0098] In this invention, external knowledge refers to knowledge that is necessary for the algorithm to solve a problem but is not directly described in the problem. Such knowledge needs to be introduced from an external knowledge base, including formulas, theorems, common knowledge, etc. For example, for a problem about solving the area of ​​a triangle, the formula for the area of ​​a triangle needs to be obtained from an external knowledge base.

[0099] The current critical state node, based on the analysis of the input state content, identifies whether additional external knowledge is required to proceed. This identification process can be based on various techniques such as keyword analysis, semantic understanding, and question type analysis. Once it is determined that external knowledge needs to be added, the current critical state node will automatically initiate the external knowledge addition method associated with the current state node.

[0100] To further illustrate the technical solution of the present invention, the following description is provided in conjunction with more specific embodiments.

[0101] Figure 2 is a schematic diagram of constructing a static graph of the solution technology using two algorithm state transition diagrams provided in an embodiment of the present invention. Referring to Figure 2, the establishment of the static graph of the solution technology can be divided into the following steps.

[0102] Step 1: Based on the existing solution algorithm A, construct state transition diagram A to represent the solution process of the existing solution algorithm A; based on the existing solution algorithm B, construct state transition diagram B to represent the solution process of the existing solution algorithm B.

[0103] The state transition diagram A includes: input node S00 Output node S os and key state node S 05 S 03 and S 02 and state transition path T 15 ,T 14 ,T 04 And T 05 .

[0104] The state transition diagram B includes: input node S 00 Output node S os and key state node S 10 S 03 and S 02 and state transition path T 16 ,T 17 ,T 04 And T 06 .

[0105] Among them, the existing solution algorithm A comes from the paper: Yu X, Lyu X, Peng R, et al. Solving arithmetic word problems by synergizing syntax-semantics extractor for explicit relations and neural network miner for implicit relations[J]. Complex & Intelligent Systems, 2022, 9(1): 697–717.

[0106] The existing solution algorithm B comes from the paper: Huang Litian, Yu Xinguo, Niu Lei, Feng Zihan. Solving algebraic problems with geometry diagrams using syntax-semantics diagram understanding[J], Computers, Materials & Continua, 2023.

[0107] It should be noted that constructing state transition diagrams based on existing solution algorithms falls under the scope of the state transition theory described above. This invention can utilize the aforementioned theory to construct state transition diagrams for most existing solution algorithms.

[0108] The focus of this invention is to construct a configuration integrated state transition graph based on multiple state transition graphs, which serves as a human-like solution engine. The construction of the state transition graph is prior art and will not be elaborated upon in this embodiment.

[0109] Step 2: Connect the input nodes S in state transition diagrams A and B. 00 Output node S os and the same key state node S 02 and S 03 Each state transition path is merged separately, and identical state transition paths are merged and integrated to form an integrated state transition diagram.

[0110] Step 3: Configure the integrated state transition diagram, including:

[0111] First, configure the diverging exit node of the integrated state transition diagram, specifically for the diverging exit node S. 00 , for S 00 Establish decision-maker D 00 D 00 Based on the content analysis of algebraic problems, the next path label for the current state node in the text content of algebraic problems is set to {S}. 05 T 15 For algebraic problems with graphical representations, the next path label for the current state node is set to {S}. 10 T 16}; For the diverging output node S 02 , for S 02 Establish decision-maker D 02 D 02 Based on the state content generated through state transition received by the current node, set the next path label as {S}. os T 05} and {S os T 06 Configure the remaining key state nodes S 03 S 05 S 10 Since each of these nodes has only one state transition path to the next node, the state transition is performed using only one state transition path.

[0112] Second, configure the convergence node S in the integrated state transition diagram. 03 The system is configured to perform data integration processing on the relational content obtained from text and the relational content obtained from graphics.

[0113] Third, configure the output node S in the integrated state transition diagram. os Based on the state content received from the state transition node, output the solution to the algebraic problem.

[0114] The above-mentioned static diagram of the solution technology indicates a simple example. In fact, the number of state transition diagrams contained in the static diagram of the solution technology is often more than two. Figure 3 is a static diagram of the solution technology provided by the embodiment of the present invention. Here, we will not go into the details of its construction process, but will focus on a brief description of its structure.

[0115] S 00 To represent the input node, it is used to receive algebraic problems, parse the content of the algebraic problems, determine the next state transition path, and perform state transition; S 01 To S 08 This represents a critical state node, used to determine the next state transition path based on the state content generated through state transitions received by the current node.

[0116] Specifically, D 00 To D 08 Indicates with S 01 To S 08 The corresponding decision-maker in this invention executes the decision-making steps described above, namely, "determining the next state transition path to perform a state transition".

[0117] S os This represents the output node, which is used to output the solution to an algebraic problem based on the state content input by the received state transition node.

[0118] T 01 ,T 02 ,T 03 ,...,T 14 , representing the 14 state transition techniques (state transition paths) in this invention.

[0119] The following explanation uses a specific algebraic problem as an example. Figure 4 is a schematic diagram of the process of solving algebraic problems using a humanoid problem-solving engine provided in this embodiment of the invention. After the problem is input into the humanoid problem-solving engine, the engine autonomously executes the humanoid problem-solving process, as follows:

[0120] Input the algebraic problem into the input node S of the static diagram of the solution technique. 00 S 00 The question content is received and the received content is input into the decision-maker D. 00 Decision Maker D 00 The decision-making process is executed through a decision model. Since the input questions typically contain both explicit and implicit relationships, the decision-maker D... 00 If the current node should transition to the vector text state, then the next path label for the current node is output as {S}. 04 T 07}, then execute the transfer technique T 07 S 00 Transition to the next critical state node S 04 ;

[0121] S 04 It is not a convergence node and does not require the execution of a content convergence process (i.e., when a critical state node receives state content transferred from multiple state transition paths, it does not need to perform data integration processing on all state content). 04 As the diverging output node, it inputs the received content into the decision-maker D. 04 Decision Maker D 04 The decision-making process is executed through a decision model, and the next path label is output as {S}. 05 T 12} and {S 06 T 13}, then execute the transfer technique T 12 S 04 Transition to the next state node S 05 Execution transfer technology T 13 S 04 Transition to the next state node S 06 In this embodiment, the transfer technology T 13 If the tags relating the distance formula and the cylinder surface area formula are identified during execution, then the distance formula "L=vt" and the cylinder surface area formula "S=πdh" need to be added to S from the knowledge base. 06 ;

[0122] S 05 and S 06 None of them are aggregation nodes, so there is no need to perform the content aggregation process, and S 05 and S 06 Neither of them are diverging outgoing nodes, S 05 The next path is {S} 03 T 14}, then execute the transfer technique T 14 S 05 Transition to the next critical state node S 03 S 06 The next path label is {S 03 T 14}, then execute the transfer technique T 14 S 05 Transition to the next critical state node S 03 ;

[0123] S 03 It is an aggregation node and needs to execute the content aggregation process. 03The received content is input into the content aggregator module. If the aggregator detects it as relational input, it initiates the cleaning and merging procedures. The cleaning procedure removes redundant relation groups based on the problem's solution intent. The merging procedure combines relations of different forms into the same representation and connects entities of the same type within the relations, thus outputting the complete relation group to S. 03 .

[0124] S 03 The aggregated data is then input into decision-maker D. 03 Decision Maker D 03 The problem involves executing a decision-making process using a decision model, outputting the label of the next path for the current state node as {S}. 02 T 04}, then execute the transfer technique T 04 S 03 Transition to the next state node S 02 ;

[0125] S 02 The received content is input into decision-maker D. 02 Decision Maker D 02 The problem involves executing a decision-making process using a decision model, outputting the next path label for the current state node as {So, T}. 06}, then execute the transfer technique T 06 S 02 Proceed to the next state node, Sos.

[0126] Based on the above steps, the solution to the algebraic problem can be achieved.

[0127] Figure 5 is a schematic diagram of the structure of the device for constructing a humanoid solution engine provided in an embodiment of the present invention. As shown in Figure 5, the device includes: a state transition diagram acquisition module 501, an integrated state transition diagram construction module 502, and a humanoid solution engine construction module 503.

[0128] The state transition diagram acquisition module 501 is used to construct a state transition diagram for each solution algorithm based on the input state, key state, output state, and the transition relationship between states constructed using transition technology, so as to represent the solution process of each solution algorithm. The state transition diagram includes input nodes, key state nodes, output nodes, and state transition paths between nodes. Input nodes correspond to input states, key state nodes correspond one-to-one with key states, output nodes correspond to output states, and state transition paths correspond one-to-one with transition relationships.

[0129] The integrated state transition diagram construction module 502 is used to merge and integrate the input nodes, output nodes, and identical key state nodes and identical state transition paths in all state transition diagrams to form an integrated state transition diagram.

[0130] The humanoid solution engine construction module 503 is used to configure the integrated state transition diagram to construct a static diagram of the solution technology as a humanoid solution engine; wherein, configuring the integrated state transition diagram includes:

[0131] Configure the input nodes in the integrated state transition diagram to receive algebraic problems, parse their content, and determine the next state transition path for state transition; configure the key state nodes in the integrated state transition diagram to determine the next state transition path based on the state content generated by the current node through state transition; configure the output nodes in the integrated state transition diagram to output the solution to the algebraic problems based on the state content received from the state transition nodes.

[0132] It should be noted that the apparatus for constructing a humanoid solution engine provided in this embodiment of the invention can execute the method for constructing a humanoid solution engine described in any of the above embodiments during specific operation, and will not be elaborated on in this embodiment.

[0133] Figure 6 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. As shown in Figure 6, the electronic device may include: a processor 610, a communication interface 620, a memory 630, and a communication bus 640. The processor 610, communication interface 620, and memory 630 communicate with each other via the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute a method for constructing a humanoid solution engine.

[0134] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the method for constructing a humanoid solution engine provided in the above embodiments.

[0135] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing a human-like solution engine, characterized in that, include: Obtain the state transition diagram for each solution algorithm; the state transition diagram includes input nodes, key state nodes, output nodes, and state transition paths between nodes. Input nodes correspond to the input states of the solution algorithm, key state nodes correspond one-to-one with the key states of the solution algorithm, output nodes correspond to the output states of the solution algorithm, and state transition paths correspond one-to-one with the transition relationships between states. The input nodes, output nodes, and identical key state nodes and identical state transition paths in all state transition diagrams are merged and integrated to form an integrated state transition diagram. The integrated state transition diagram is configured to construct a static diagram of the solution technology as a human-like solution engine; wherein, the method for configuring the integrated state transition diagram includes: Configure the input nodes in the integrated state transition diagram to receive algebraic problems, parse their content, and determine the next state transition path for state transition; configure the key state nodes in the integrated state transition diagram to determine the next state transition path based on the state content generated by the current node through state transition; configure the output nodes in the integrated state transition diagram to output the solution to the algebraic problems based on the state content received from the state transition nodes. Specifically, for input nodes and key state nodes, determining the next state transition path for state transition includes: When a node has multiple state transition paths, at least one target state transition path is determined from the multiple state transition paths based on the currently received state content to perform a state transition. When a node has only one state transition path, the state transition is performed using only that one state transition path. When a node has multiple state transition paths, the node is referred to as a diverging transition node; based on the currently received state content, at least one target state transition path is determined from the multiple transition paths to perform a state transition, including: The currently received state content is input into the first decision model of the pre-built current divergent outgoing node, and the predicted path label is output for state transition. The steps for constructing the first decision model for any divergent outgoing node include: Obtain the decision dataset of any diverging outgoing node regarding the state transition path; the decision dataset includes the state content received by any diverging outgoing node and the corresponding path label; wherein, the path label includes the next state transition path and the label of the node; Based on the aforementioned decision dataset, the rule-driven mapping relationship between state content and path labels is analyzed to construct a first decision model; When a node has multiple state transition paths, the node is referred to as a diverging transition node; based on the currently received state content, at least one target state transition path is determined from the multiple transition paths to perform a state transition, including: The currently received state content is input into the pre-trained second decision model of the current divergent transition node, and the predicted path label is output for state transition. The steps for constructing the second decision model for any divergent outgoing node include: Obtain the decision dataset of any diverging outgoing node regarding the state transition path; the decision dataset includes the state content received by any diverging outgoing node and the corresponding path label; wherein, the path label includes the next state transition path and the label of the node; A pre-defined machine learning model is trained using a decision dataset, and the trained pre-defined machine learning model is used as the second decision model. Also includes: When the key state node receives state content transferred from multiple state transition paths, it performs data integration processing on all state content.

2. The method for constructing a humanoid solution engine according to claim 1, characterized in that, Also includes: If necessary knowledge is lacking during state transition, the required knowledge can be obtained from a pre-set knowledge base.

3. The method for constructing a humanoid solution engine according to claim 1, characterized in that, Also includes: When a new solution algorithm emerges, the configuration of the humanoid solution engine is updated.

4. The method for constructing a humanoid solution engine according to claim 1, characterized in that, The data integration processing methods for all status content include one or more of the following: information merging, content cleaning, and priority sorting.

5. An apparatus for constructing a human-like solution engine, characterized in that, include: The state transition diagram acquisition module is used to acquire the state transition diagram for each solution algorithm; The state transition diagram includes input nodes, key state nodes, output nodes, and state transition paths between nodes. Input nodes correspond to the input states of the solution algorithm, key state nodes correspond one-to-one with the key states of the solution algorithm, output nodes correspond to the output states of the solution algorithm, and state transition paths correspond one-to-one with the transition relationships between states. The integrated state transition diagram construction module is used to merge and integrate the input nodes, output nodes, and identical key state nodes and identical state transition paths in all state transition diagrams to form an integrated state transition diagram. A humanoid solution engine construction module is used to configure the integrated state transition diagram to construct a static diagram of the solution technology as a humanoid solution engine; wherein, configuring the integrated state transition diagram includes: Configure the input nodes in the integrated state transition diagram to receive algebraic problems, parse their content, and determine the next state transition path for state transition; configure the key state nodes in the integrated state transition diagram to determine the next state transition path based on the state content generated by the current node through state transition; configure the output nodes in the integrated state transition diagram to output the solution to the algebraic problems based on the state content received from the state transition nodes. Specifically, for input nodes and key state nodes, determining the next state transition path for state transition includes: When a node has multiple state transition paths, at least one target state transition path is determined from the multiple state transition paths based on the currently received state content to perform a state transition. When a node has only one state transition path, the state transition is performed using only that one state transition path. When a node has multiple state transition paths, the node is referred to as a diverging transition node; based on the currently received state content, at least one target state transition path is determined from the multiple transition paths to perform a state transition, including: The currently received state content is input into the first decision model of the pre-built current divergent outgoing node, and the predicted path label is output for state transition. The steps for constructing the first decision model for any divergent outgoing node include: Obtain the decision dataset of any diverging outgoing node regarding the state transition path; the decision dataset includes the state content received by any diverging outgoing node and the corresponding path label; wherein, the path label includes the next state transition path and the label of the node; Based on the aforementioned decision dataset, the rule-driven mapping relationship between state content and path labels is analyzed to construct a first decision model; When a node has multiple state transition paths, the node is referred to as a diverging transition node; based on the currently received state content, at least one target state transition path is determined from the multiple transition paths to perform a state transition, including: The currently received state content is input into the pre-trained second decision model of the current divergent transition node, and the predicted path label is output for state transition. The steps for constructing the second decision model for any divergent outgoing node include: Obtain the decision dataset of any diverging outgoing node regarding the state transition path; the decision dataset includes the state content received by any diverging outgoing node and the corresponding path label; wherein, the path label includes the next state transition path and the label of the node; A pre-defined machine learning model is trained using a decision dataset, and the trained pre-defined machine learning model is used as the second decision model. Among them, the key state node is configured to perform data integration processing on all state contents when the key state node receives state contents transferred by multiple state transition paths.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for constructing a humanoid solution engine as described in any one of claims 1 to 4.

Images