English word memory system
By constructing a knowledge graph semantic network and dynamically generating multi-level contexts, combined with personalized learning path planning, the problems of high forgetting rate and low learning efficiency in the existing English word memory system are solved, and more efficient and personalized learning effects are achieved.
Patent Information
- Application Number
- CN202510037079.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-09
- Publication Date
- 2025-05-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing English word memory system lacks semantic association and context support, resulting in high user forgetting rate, low learning efficiency, and unscientific planning of personalized learning paths.
By building a semantic network based on knowledge graphs, a multi-level context of the target words is dynamically generated, including phrases, sentences, paragraphs and scene-based chapters, and a personalized learning path is planned based on the semantic network weight distribution and user learning records.
It improves the efficiency and interest of word memory, significantly reduces the forgetting rate, and makes the learning content more scientific and targeted through personalized learning paths.
Smart Images
Figure CN119940508A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of memory assistance, in particular to an English word memory system. Background Art
[0002] Most current English word memory systems use a memory method based on mechanical repetition and word lists. Although it can improve the amount of memory in a short period of time, it is easy to lead to high forgetting rate and low learning efficiency of users. This method ignores the usage characteristics of words in actual contexts and lacks semantic associations and contextual support, which makes it difficult for learning content to be integrated into users' long-term memory. In addition, some memory systems attempt to optimize word memory through image association, context examples, etc., but they are often limited to static context generation, and fail to dynamically adjust the content according to users' learning habits and memory characteristics, nor do they achieve in-depth mining and utilization of semantic relationships between words.
[0003] Especially in terms of personalized learning path planning, traditional methods often adopt a fixed order or user-defined method, failing to consider the semantic association, learning frequency and forgetting rules of words, resulting in the lack of scientific and targeted organization of learning content. In addition, existing systems mostly present learning content in a single mode, such as only providing word definitions and static examples, while ignoring the role of multi-level contexts (such as phrases, sentences, paragraphs, and scenario-based chapters) in strengthening users' word application ability. Summary of the invention
[0004] The invention provides an English word memory system.
[0005] An English word memory system, comprising:
[0006] Semantic network construction module: Use knowledge graph technology and word vector model to construct a semantic network topology structure. The semantic network includes nodes and edges. Nodes represent words, and edges represent semantic relationships between words. Semantic relationships include synonyms, antonyms, hyponyms, and near synonyms. By calculating the semantic similarity, contextual relevance, and co-occurrence frequency between words, edge weights between words are generated. The edge weights represent the strength of the semantic relationship. The semantic network is represented by a weighted directed graph.
[0007] Dynamic context generation module: Based on the node and edge weights in the semantic network, the module dynamically generates a multi-level context of the target word input by the user through natural language generation technology, including phrases, sentences, paragraphs and scenario chapters;
[0008] Dynamic memory path planning module: plans personalized learning paths based on the multi-level context data and weight distribution of the semantic network provided by the dynamic context generation module; the planning of the personalized learning path comprehensively considers the semantic relationship of the target words, the user's learning frequency and forgetting rate, analyzes the context adaptability, gives priority to target words and their associated words with high context adaptability, and aligns and displays the multi-level context generated by the dynamic context generation module with the words and their associated words sorted by the personalized learning path.
[0009] Optionally, the semantic network construction module specifically includes:
[0010] Node generation: Extract words and their semantic relationships from a predefined vocabulary database (WordNet) and corpus (COCA) using knowledge graph technology, including synonyms, antonyms, hyponyms, and near synonyms. Each word and its related words are used as nodes. Node attributes include the word's part of speech (such as noun, verb), semantic category (such as abstract concept, concrete object), and frequency of use.
[0011] Edge generation and semantic similarity calculation: The semantic similarity between words is calculated through the word vector model, including semantic similarity calculation based on semantic vectors; for hyponym and hyponym relationships, the context relevance is calculated through the shortest distance of the knowledge graph path; for words that appear in the same context, the co-occurrence frequency of words in the same paragraph or sentence is measured in combination with corpus statistical data;
[0012] Generation of edge weights and construction of weighted directed graphs: According to semantic similarity, contextual relevance, and co-occurrence frequency, edge weights between words are generated. The constructed semantic network is represented in the form of a weighted directed graph, where nodes represent words, edges represent semantic relationships, and edge weights represent relationship strength.
[0013] Optionally, the semantic similarity SemanticSim i,j Cosine similarity calculation based on word vector model:
[0014] in, is the word vector representation of words i and j, is the word vector The module is expressed as where v i,k is the k-th dimension component of the word vector, is the word vector Model.
[0015] Optionally, the contextual correlation ContextualSim i,jMeasures the strength of the association between two words i and j in the knowledge graph, based on the shortest path distance calculation of word nodes in the knowledge graph: ContextualSi Among them, PathLength i,j Indicates the shortest path length between word i and j nodes in the knowledge graph. The smaller the value, the closer the association. i,j The closer to 1.
[0016] Optionally, the co-occurrence frequency Frequency i,j A combination of corpus-based statistics to measure the strength of the frequency of occurrence between words: Among them, CoOccurrence i,j TotalOccurrences indicates the number of times words i and j co-occur in the same paragraph or sentence in the corpus. i,j Represents the maximum number of co-occurrences of all word pairs in the corpus.
[0017] Optionally, the edge weight is expressed as:
[0018] W i,j =α·SemanticSim i,j +β·ContextualSim i,j +γ·Frequency i,j ,That
[0019] In , α, β, and γ are weight parameters, which control the influence of semantic similarity, contextual relevance, and co-occurrence frequency respectively;
[0020] SemanticSim i,j 、ContextualSim i,j Frequency i,j They represent the semantic similarity, contextual relevance and co-occurrence frequency of word pairs respectively, and i,j represent word i and word j respectively.
[0021] Optionally, the dynamic context generation module specifically includes:
[0022] Receive the target word set T specified by the user = {t1, t2, ..., t n}, extract the nodes (associated words) and edge weights associated with each target word from the semantic network. The nodes represent semantic associated words, and the edge weights represent the strength of the semantic relationship. Nodes with high edge weights in the semantic network are selected first to form the core semantic vocabulary set C = {c1, c2, ..., c m};
[0023] Multi-level context generation, including:
[0024] Phrase generation: Based on the core semantic vocabulary set C, a fixed phrase template ([target word] + [related word]) is used to generate phrases;
[0025] Sentence generation: Generate grammatically complete sentences by combining target words and semantically related words through natural language generation technology;
[0026] Paragraph generation: Based on phrases and sentences, multiple related words are combined to generate logically coherent paragraphs, ensuring that the paragraph content conforms to the lexical relationships in the semantic network;
[0027] Scenario-based chapter generation: Generate chapter content in specific scenarios for target words through context enhancement mechanism;
[0028] Output contextual content: Output generated phrases, sentences, paragraphs, and scenario-based chapters as learning content for users.
[0029] Optionally, the dynamic memory path planning module specifically includes:
[0030] Input data: including multi-level context data, semantic network weights, and the user's learning frequency and forgetting rate. The semantic network weights extract the edge weights of the target word and its associated words from the semantic network, indicating the strength of the semantic relationship.
[0031] Path priority calculation: According to the target word t i The semantic network weight distribution is used to calculate the contextual adaptability score S of the target word in the multi-level context. context , and comprehensively consider the learning frequency F of the target word i and the forgetting rate U i , calculate the comprehensive priority score P i ;
[0032] Learning path planning: Based on the comprehensive priority score P i , sort the target words and their related words, give priority to high-priority words, and build a personalized learning path L = {t1, t2, ..., t n}, words with high context adaptability and their associated words are presented first in the path.
[0033] Optionally, the context suitability score is calculated as: Among them, C i is the target word t i The associated word set, W i,j is the target word t i and the conjunction t j The edge weights of the semantic network;
[0034] The composite priority score is calculated as: Among them, F i is the learning frequency of the target word, and its value range is [0,1]. The higher the value, the more times the user has learned the word. i represents the forgetting rate of the target word, which is dynamically calculated based on the user's learning record. They are weight coefficients, which are used to adjust the impact of context adaptability, learning frequency and forgetting rate on priority.
[0035] Optionally, the aligned presentation includes taking the priority of each target word and its associated words in the personalized learning path as the presentation order of the multi-level contextual content:
[0036] Prioritize the presentation of phrases and sentences directly related to high-priority target words;
[0037] After the user has learned the phrases and sentences, the paragraph content is gradually presented;
[0038] Finally, a scenario-based chapter is displayed to allow users to combine words with specific scenarios.
[0039] Beneficial effects of the present invention:
[0040] The present invention dynamically generates a multi-level context of the target word, including phrases, sentences, paragraphs and scenario-based chapters, by constructing a semantic network based on the knowledge graph. Compared with the traditional isolated memory method, the dynamic context generation module ensures that the learning process of each word is closely integrated with its semantic association, so that users can better understand and apply words in the real context. For example, through the step-by-step guidance of phrases and sentences, users can quickly master the core usage of words, while the content of paragraphs and chapters further strengthens the memory and application ability of words in complex situations, significantly improving memory efficiency and learning interest.
[0041] The present invention adopts a dynamic memory path planning module, and dynamically plans a learning path with the goal of minimizing the forgetting rate by combining the semantic network weight distribution, multi-level context data and user learning records. The personalized path dynamically adjusts the priority according to the semantic association, learning frequency and forgetting rate of the target word, ensuring that the user has priority in learning the word content with high context adaptability and close semantic association. The dynamic calculation of the forgetting rate and learning frequency is introduced, so that the system can adapt to the user's memory characteristics, greatly reduce the forgetting rate, and improve the organization efficiency of the learning content.
[0042] The present invention closely combines the personalized learning path with the dynamic context generation module, so that the word order of the learning path is linked with the multi-level context content. The learning process goes from phrases to sentences, paragraphs and then to chapters, deepening the semantics and application understanding of words layer by layer. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0044] Figure 1 A schematic diagram of a memory system functional module according to an embodiment of the present invention;
[0045] Figure 2 A schematic diagram of generating a personalized learning path according to an embodiment of the present invention. DETAILED DESCRIPTION
[0046] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.
[0047] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).
[0048] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.
[0049] like Figure 1-Figure 2 As shown, an English word memory system comprises:
[0050] Semantic network construction module: Use knowledge graph technology and word vector model to build semantic network topology. The semantic network includes nodes and edges. Nodes represent words, and edges represent semantic relationships between words. Semantic relationships include synonyms, antonyms, hyponyms, and near synonyms. By calculating the semantic similarity, contextual relevance, and co-occurrence frequency between words, edge weights between words are generated. Edge weights represent the strength of semantic relationships. The semantic network is represented by a weighted directed graph.
[0051] Dynamic context generation module: Based on the node and edge weights in the semantic network, the module dynamically generates a multi-level context of the target word input by the user through natural language generation technology, including phrases, sentences, paragraphs and scenario chapters;
[0052] Dynamic memory path planning module: plans personalized learning paths based on the multi-level context data and weight distribution of the semantic network provided by the dynamic context generation module; the planning of personalized learning paths comprehensively considers the semantic relationship of the target words, the user's learning frequency and forgetting rate, analyzes the context adaptability, gives priority to target words and their associated words with high context adaptability, and aligns and displays the multi-level context generated by the dynamic context generation module with the words and their associated words sorted by the personalized learning path.
[0053] The semantic network building blocks include:
[0054] Node generation: Extract words and their semantic relationships from a predefined vocabulary database (WordNet) and corpus (COCA) using knowledge graph technology, including synonyms, antonyms, hyponyms, and near synonyms. Each word and its related words are used as nodes. Node attributes include the word's part of speech (such as noun, verb), semantic category (such as abstract concept, concrete object), and frequency of use.
[0055] Edge generation and semantic similarity calculation: The semantic similarity between words is calculated through the word vector model, including semantic similarity calculation based on semantic vectors; for hyponym and hyponym relationships, the context relevance is calculated through the shortest distance of the knowledge graph path; for words that appear in the same context, the co-occurrence frequency of words in the same paragraph or sentence is measured in combination with corpus statistical data;
[0056] Generation of edge weights and construction of weighted directed graphs: According to semantic similarity, contextual relevance, and co-occurrence frequency, edge weights between words are generated. The constructed semantic network is represented in the form of a weighted directed graph, where nodes represent words, edges represent semantic relationships, and edge weights represent relationship strength.
[0057] SemanticSim i,j Cosine similarity calculation based on word vector model:
[0058] in, is the word vector representation of words i and j, is the word vector The module is expressed as where v i,k is the k-th dimension component of the word vector, is the word vector Model.
[0059] ContextualSim i,j Measures the strength of association between two words i and j in the knowledge graph, based on the shortest path distance calculation of word nodes in the knowledge graph: Among them, PathLength i,j Indicates the shortest path length between word i and j nodes in the knowledge graph. The smaller the value, the closer the association. i,j The closer to 1; the knowledge graph constructs words into a network through semantic relationships, in which the shortest distance of the path represents the strength of association between two node words. The hyponym and hyponym relationship (such as "animal" → "dog") represents a hierarchical association from general to specific. They often appear as progressive or inclusive relationships in the context. The path length of the knowledge graph can directly reflect the hierarchical span between two words (the shorter the distance, the stronger the association). The smaller the shortest path length, it means that the two words are connected by a more direct and relevant semantic relationship, and therefore are more likely to be associated with each other in the context.
[0060] Frequency i,j A combination of corpus-based statistics to measure the strength of the frequency of occurrence between words: Among them, CoOccurrence i,j TotalOccurrences indicates the number of times words i and j co-occur in the same paragraph or sentence in the corpus. i,j Represents the maximum number of co-occurrences of all word pairs in the corpus.
[0061] The edge weight is expressed as:
[0062] Among them, α, β, and γ are weight parameters, which control the influence of semantic similarity, contextual relevance, and co-occurrence frequency respectively;
[0063] SemanticSim i,j 、ContextualSim i,j Frequency i,jThey represent the semantic similarity, contextual relevance and co-occurrence frequency of word pairs respectively, and i,j represent word i and word j respectively.
[0064] The dynamic context generation module specifically includes:
[0065] Receive the target word set T specified by the user = {t1, t2, ..., t n}, extract the nodes (associated words) and edge weights associated with each target word from the semantic network. The nodes represent semantic associated words, and the edge weights represent the strength of the semantic relationship. Nodes with high edge weights in the semantic network are selected first to form the core semantic vocabulary set C = {c1, c2, ..., c m};
[0066] Multi-level context generation, including:
[0067] Phrase generation: Based on the core semantic vocabulary set C, a fixed phrase template ([target word] + [associated word]) is used to generate phrases. For example, if the target word is "run" and the associated word is "fast", the phrase "run fast" is generated.
[0068] Sentence generation: Using natural language generation technology, we combine target words and semantically related words to generate grammatically complete sentences, such as “The athlete can run fast to win the race”.
[0069] Paragraph generation: Based on phrases and sentences, combine multiple related words to generate logically coherent paragraphs, ensuring that the paragraph content conforms to the lexical relationship in the semantic network. For example, a paragraph describing a sports scene contains words such as "run", "race", and "athlete".
[0070] Scenario-based paragraph generation: Generate paragraph content for target words in specific scenarios through context enhancement mechanism, such as generating a paragraph describing a track and field competition in the "sports competition" scenario, combining semantically related words such as "run", "sprint", and "finishline";
[0071] Output contextual content: Output generated phrases, sentences, paragraphs, and scenario-based chapters as learning content for users.
[0072] The specific method of natural language generation technology combining target words and associated words to generate grammatically complete sentences is as follows:
[0073] 1. Syntactic template matching:
[0074] Design a variety of standard grammar templates in advance to combine target words and their associated words. For example:
[0075] Template 1 (subject-verb-object structure): [subject][verb][object]
[0076] Template 2 (time adverbial structure): [time adverbial], [subject] [predicate]
[0077] Example template:
[0078] [target word] can [related verb] [object] in [related noun]
[0079] [target word] is used to [related verb] in [related context]
[0080] 2. Semantic filling:
[0081] Combined with the associated words in the semantic network, the placeholders in the template (such as [target word], [associated verb], [associated noun]) are filled with matching words.
[0082] Target word: Select from the input learning target, such as "run".
[0083] Related verbs: Extract words from the semantic network that have a strong semantic relationship with the target word, such as "sprint".
[0084] Related noun: Choose a related word, such as "race".
[0085] 3. Syntax check:
[0086] Use natural language processing tools to parse the generated sentences to ensure that they comply with syntactic rules.
[0087] Example sentence generation:
[0088] Target word: run
[0089] Template: [subject][predicate][object]in[location]
[0090] Generate the sentence: The athlete can run fast in the marathon.
[0091] The content of the chapter generated in a specific scenario through the context enhancement mechanism is as follows:
[0092] 1. Scene keyword extraction:
[0093] Determine the target scenario (such as "sports competition"), extract the core vocabulary of the scenario (such as "athlete", "sprint", "marathon"), combine with the semantic network, and add extended associated words to the target words to form a scenario vocabulary.
[0094] 2. Construction of scene chapter template:
[0095] Design a chapter template that fits the scenario, covering paragraph structure and content logic. For example:
[0096] Paragraph 1: Describe the setting of the scene.
[0097] Paragraph 2: Contextual expansion using target words and related words.
[0098] The third paragraph: summary or outlook.
[0099] 3. Language model generation: Input the target word and its related words into the pre-trained language model, specify the target scenario, combine the scenario vocabulary and template, and guide the model to generate chapter content.
[0100] 4. Generate examples:
[0101] Target word: run;
[0102] Scene: Sports competition;
[0103] Then generate chapters;
[0104] 5. Contextual relevance check: Verify whether the generated content is relevant to the target word and scenario: whether the target word "run" is used; whether the scenario-related words "marathon", "sprint", and "athlete" are included.
[0105] The dynamic memory path planning module specifically includes:
[0106] Input data: including multi-level context data, semantic network weights, and the user's learning frequency and forgetting rate. The semantic network weights extract the edge weights of the target word and its associated words from the semantic network, indicating the strength of the semantic relationship.
[0107] Path priority calculation: According to the target word t i The semantic network weight distribution is used to calculate the contextual adaptability score S of the target word in the multi-level context. context , and comprehensively consider the learning frequency F of the target word i and the forgetting rate U i , calculate the comprehensive priority score P i ;
[0108] Learning path planning: Based on the comprehensive priority score P i , sort the target words and their related words, give priority to high-priority words, and build a personalized learning path L = {t1, t2, ..., t n}, words with high context adaptability and their associated words are presented first in the path.
[0109] The contextual fit score is calculated as: Among them, C i is the target word t i The associated word set, W i,j is the target word t i and the conjunction t j The edge weights of the semantic network;
[0110] The composite priority score is calculated as: Among them, F i is the learning frequency of the target word, and its value range is [0,1]. The higher the value, the more times the user has learned the word. i represents the forgetting rate of the target word, which is dynamically calculated based on the user's learning record. They are weight coefficients, which are used to adjust the impact of context adaptability, learning frequency and forgetting rate on priority.
[0111] Learning frequency F i Used to indicate the user's response to the target word t i The normalized result of the number of learning times reflects the user's familiarity with the word. The calculation formula is: Among them, LearningCount i Represents the target word t i The cumulative number of learning times in the user’s learning record. max(LearningCount) indicates the maximum number of learning times of all target words in the user’s learning record. It is used for normalization. i The value range of F is [0,1]. i =0 means the user has not learned the word yet, F i =1 means the user has learned this word the most times.
[0112] Forgetting rate U i Used to indicate the user's response to the target word t i The degree of forgetting is calculated based on the Ebbinghaus forgetting curve and the learning time interval. The calculation formula is: Where, ΔT i is the target word t i The interval between the last learning time and the current time, T opt is the user's optimal review interval for the target word (obtained by fitting the forgetting curve), κ is the forgetting rate factor, which is set according to the complexity of the word. The larger the value, the faster the forgetting. i is the user's memory accuracy of the target word, with a value range of [0,1]. Among them, CorrectAttempts i Indicates the user's understanding of word t during the learning process iThe number of correct answers, TotalAttempts i Indicates the user's response to word t i The total number of answers, U i The value range of U is [0,1]. i =0 means that the user has a very strong memory of the word, U i =1 means the user has completely forgotten the word.
[0113] From the multi-level context content output by the dynamic context generation module, extract the corresponding content of each target word and its associated words in the phrases, sentences, paragraphs and scenario chapters of the multi-level context. Aligned display includes taking the priority of each target word and its associated words in the personalized learning path as the presentation order of the multi-level context content, as follows:
[0114] 1. Prioritize the display of phrases and sentences directly related to high-priority target words;
[0115] 2. After the user has completed the learning of phrases and sentences, the paragraph content is gradually presented;
[0116] 3. Finally, a scenario-based chapter is presented to enable users to combine words with specific scenarios and enhance their application capabilities.
[0117] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.
[0118] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. An English word memory system, characterized in that: include: Semantic network construction module: Use knowledge graph technology and word vector model to construct a semantic network, which includes nodes and edges. Nodes represent words, and edges represent semantic relationships between words. Semantic relationships include synonyms, antonyms, hyponyms, and near synonyms. By calculating the semantic similarity, contextual relevance, and co-occurrence frequency between words, edge weights between words are generated. The edge weights represent the strength of the semantic relationship. The semantic network is represented by a weighted directed graph. Dynamic context generation module: Based on the node and edge weights in the semantic network, the module dynamically generates a multi-level context of the target word input by the user through natural language generation technology, including phrases, sentences, paragraphs and scenario chapters; Dynamic memory path planning module: plans a personalized learning path based on the multi-level context data and weight distribution of the semantic network provided by the dynamic context generation module; the planning of the personalized learning path comprehensively considers the semantic relationship, learning frequency and forgetting rate of the target words, analyzes the context adaptability, gives priority to the target words and their associated words with high context adaptability, and aligns and displays the multi-level context generated by the dynamic context generation module with the words and their associated words sorted by the personalized learning path.
2. An English word memory system according to claim 1, characterized in that: The semantic network building module specifically includes: Node generation: Extract words and their semantic relationships from predefined vocabulary databases and corpora through knowledge graph technology, including synonyms, antonyms, hyponyms, and near synonyms. Each word and its related words are used as nodes. Node attributes include the word's part of speech, semantic category, and frequency of use. Edge generation and semantic similarity calculation: The semantic similarity between words is calculated through the word vector model, including semantic similarity calculation based on semantic vectors; for hyponym and hyponym relationships, the context relevance is calculated through the shortest distance of the knowledge graph path; for words that appear in the same context, the co-occurrence frequency of words in the same paragraph or sentence is measured in combination with corpus statistical data; Generation of edge weights and construction of weighted directed graphs: According to semantic similarity, contextual relevance, and co-occurrence frequency, edge weights between words are generated. The constructed semantic network is represented in the form of a weighted directed graph, where nodes represent words, edges represent semantic relationships, and edge weights represent relationship strength.
3. An English word memory system according to claim 2, characterized in that: The semantic similarity SemanticSim i,j Cosine similarity calculation based on word vector model: in, is the word vector representation of words i and j, is the word vector The module is expressed as where v i,k is the k-th dimension component of the word vector, is the word vector Model.
4. The English word memory system according to claim 3, characterized in that: The contextual relevance of ContextualSim i,j Measures the strength of association between two words i and j in the knowledge graph, based on the shortest path distance calculation of word nodes in the knowledge graph: Among them, PathLength i,j Indicates the shortest path length between word i and j nodes in the knowledge graph. The smaller the value, the closer the association. i,j The closer to 1.
5. An English word memory system according to claim 4, characterized in that: The co-occurrence frequency i,j A combination of corpus-based statistics to measure the strength of the frequency of occurrence between words: Among them, CoOccurrence i,j TotalOccurrences indicates the number of times words i and j co-occur in the same paragraph or sentence in the corpus. i,j Represents the maximum number of co-occurrences of all word pairs in the corpus.
6. An English word memory system according to claim 5, characterized in that: The edge weight is expressed as: Among them, α, β, and γ are weight parameters, which control the influence of semantic similarity, contextual relevance, and co-occurrence frequency respectively; SemanticSim i,j 、ContextualSim i,j Frequency i,j They represent the semantic similarity, contextual relevance and co-occurrence frequency of word pairs respectively, and i,j represent word i and word j respectively.
7. The English word memory system according to claim 1, characterized in that: The dynamic context generation module specifically includes: Receive the target word set T specified by the user = {t1, t2, ..., t n }, extract the nodes and edge weights associated with each target word from the semantic network. The nodes represent semantically related words, and the edge weights represent the strength of the semantic relationship. Nodes with high edge weights in the semantic network are selected first to form the core semantic vocabulary set C = {c1, c2, ..., c m }; Multi-level context generation, including: Phrase generation: Based on the core semantic vocabulary set C, a fixed phrase template ([target word] + [related word]) is used to generate phrases; Sentence generation: Generate grammatically complete sentences by combining target words and semantically related words through natural language generation technology; Paragraph generation: Based on phrases and sentences, multiple related words are combined to generate logically coherent paragraphs, ensuring that the paragraph content conforms to the lexical relationships in the semantic network; Scenario-based chapter generation: Generate chapter content in specific scenarios for target words through context enhancement mechanism; Output contextual content: Output generated phrases, sentences, paragraphs, and scenario-based chapters as learning content for users.
8. The English word memory system according to claim 1, characterized in that: The dynamic memory path planning module specifically includes: Input data: including multi-level context data, semantic network weights, and users’ learning frequency and forgetting rate; Path priority calculation: According to the target word t i The semantic network weight distribution is used to calculate the contextual adaptability score S of the target word in the multi-level context. context , and comprehensively consider the learning frequency F of the target word i and the forgetting rate U i , calculate the comprehensive priority score P i ; Learning path planning: Based on the comprehensive priority score P i , sort the target words and their related words, give priority to high-priority words, and build a personalized learning path L = {t1, t2, ..., t n }, words with high context adaptability and their associated words are presented first in the path.
9. An English word memory system according to claim 8, characterized in that: The context suitability score is calculated as: Among them, C i is the target word t i The associated word set, W i,j is the target word t i and the conjunction t j The edge weights of the semantic network; The composite priority score is calculated as: Among them, F i is the learning frequency of the target word, ranging from [0,1], U i represents the forgetting rate of the target word, δ,θ, They are weight coefficients, which are used to adjust the impact of context adaptability, learning frequency and forgetting rate on priority.
10. The English word memory system according to claim 1, characterized in that: The alignment presentation includes taking the priority of each target word and its associated words in the personalized learning path as the presentation order of the multi-level contextual content: Prioritize the presentation of phrases and sentences directly related to high-priority target words; After the user has learned the phrases and sentences, the paragraph content is gradually presented; Finally, a scenario-based chapter is displayed to allow users to combine words with specific scenarios.