Auxiliary memory method and system based on machine learning

Abstract

The invention relates to an auxiliary memory method and system based on machine learning, and belongs to the field of machine learning, and the method comprises the steps: setting the length of a user input bit according to the length of a target memory word letter sequence; obtaining a memory feedback text input by the user; the Hamming distance between the target memory word letter sequence and the memory feedback letter sequence is calculated, and if the Hamming distance is larger than 0, the corresponding position of the target memory word letter sequence reckoned in the Hamming distance is marked as a forgetting position; and when the user interaction module displays the target memory word, the letters of the forgotten positions of the target memory word are displayed differently. According to the method and the device, the word memory effect can be distinguished more finely, the letters of the forgotten positions in the word are displayed differently in combination with the forgotten positions, the user is helped to focus on the part which is easier to forget, and the auxiliary memory effect is improved.

Description

Technical Field

[0001] This application relates to the field of machine learning technology, and in particular to a machine learning-based auxiliary memory method and system. Background Technology

[0002] Current vocabulary memorization apps typically rely on the forgetting curve, setting review intervals for target words and displaying them to the user based on these intervals to help maintain a high retention rate. However, these methods use the entire word as the unit for evaluating review and retention, which limits their ability to differentiate between different parts of a word for more detailed memorization. For example, a common scenario is that users remember certain letters of a word (such as the first or last letter) well and retain them long-term, while remembering other parts of the word poorly and forgetting them easily. Current methods cannot differentiate between different parts of a word for targeted memorization, hindering further improvement in memorization effectiveness. Summary of the Invention

[0003] To address the aforementioned issues, this application provides a machine learning-based method and system for assisted memory.

[0004] In a first aspect, this application provides a machine learning-based assisted memory method, comprising: Based on the letter sequence length of the target word, set the length of the user input field for the target word in the user interaction module; The user interaction module obtains the user's input feedback text on the target word. Based on the memory feedback text, the memory feedback letter sequence is obtained; Calculate the Hamming distance between the letter sequence of the target memorized word and the letter sequence of the memory feedback. If the Hamming distance is greater than 0, then the letter sequence of the target memorized word is included in the corresponding position of the Hamming distance and marked as the forgetting position. When the target word is displayed in the user interaction module, the letters of the forgotten position of the target word are displayed in a differentiated manner.

[0005] Furthermore, the machine learning-based memory-aided method further includes the following steps: Based on the letter sequence of the target word, its forgotten positions, and the input time of the corresponding memory feedback text, a word-assisted memory sample is established. Using multiple word-assisted memory samples, with the number of forgotten positions of the letter sequence of the same target word decreasing over time as a constraint, an objective function is set, and the machine learning model is trained to classify the increase and decrease of display frequency, thus obtaining the display frequency increase and decrease classification model. When the target word is displayed in the user interaction module, the display frequency of the target word is updated based on the display frequency increase / decrease classification model.

[0006] Furthermore, the machine learning-based memory-aided method further includes the following steps: The forgetting rate of the target word is determined based on the Hamming distance between the letter sequence of the target word and the letter sequence of the memory feedback. The frequency of display of the target words is determined based on their forgetting rate.

[0007] Further, calculating the Hamming distance between the letter sequence of the target memorized word and the letter sequence of the memory feedback includes the following steps: The letters in the target word sequence and the memory feedback sequence are unified to lowercase or uppercase. For the target word sequence and the memory feedback sequence after being unified to lowercase or uppercase, XOR calculation is performed on the letters with the same position, and the results of the XOR calculation are accumulated to obtain the Hamming distance between the target word sequence and the memory feedback sequence.

[0008] Furthermore, the machine learning-based memory-aided method further includes the following steps: If the number of word-aided memory samples does not exceed one thousand, then the machine learning model adopts the support vector machine model.

[0009] Furthermore, the differentiated display of the forgotten letters of the target memorized word includes the following steps: Highlight or enlarge the letters in the forgotten positions of the target word.

[0010] Furthermore, highlighting the forgotten letters of the target word includes the following steps: Based on the input time of the memory feedback text corresponding to the forgotten position of the target word, the letters of the forgotten position of the target word are highlighted or enlarged in a graded manner.

[0011] Secondly, this application provides a machine learning-based assisted memory system, comprising: The user interaction module is configured to: set the length of the user input position of the target word in the user interaction module according to the length of the letter sequence of the target word; obtain the user input feedback text on the target word; and when the target word is displayed in the user interaction module, update the display frequency of the target word based on the display frequency increase / decrease classification model, and differentiate the letters of the forgotten position of the target word. The data processing module is configured to: obtain a memory feedback letter sequence based on the memory feedback text; calculate the Hamming distance between the letter sequence of the target memory word and the memory feedback letter sequence; if the Hamming distance is greater than 0, mark the corresponding position of the letter sequence of the target memory word as a forgotten position in the Hamming distance; establish a word auxiliary memory sample based on the letter sequence of the target memory word, its forgotten positions, and the input time of the corresponding memory feedback text; use multiple word auxiliary memory samples, with the number of forgotten positions of the letter sequence of the same target memory word decreasing over time as a constraint, set an objective function, and train the machine learning model to perform display frequency increase / decrease classification training to obtain a display frequency increase / decrease classification model.

[0012] Furthermore, the data processing module is also configured to: determine the forgetting rate of the target word based on the Hamming distance between the letter sequence of the target word and the letter sequence of the memory feedback; and determine the display frequency of the target word based on the forgetting rate of the target word.

[0013] Furthermore, the data processing module is specifically used to: unify the letters in the target memorized word letter sequence and the memory feedback letter sequence to lowercase or uppercase; perform XOR calculation on each letter in the target memorized word letter sequence and the memory feedback letter sequence after unification to lowercase or uppercase, and accumulate the XOR calculation results to obtain the Hamming distance between the target memorized word letter sequence and the memory feedback letter sequence.

[0014] Furthermore, the machine learning model is a support vector machine model.

[0015] The beneficial effects of the above-mentioned technical solutions provided in this application include at least the following: The technical solution of this application can make more detailed distinctions in word memorization effects. By combining the forgetting points, it can differentiate and display the letters in the forgotten points of a word, helping users to focus their attention on the more easily forgotten parts and thus improving the effectiveness of assisted memorization.

[0016] Other features and advantages of this application will be set forth in the following description or may be learned by practicing this application.

[0017] The technical solution of this application will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the present application and form part of the specification. They are used together with the embodiments of the present application to explain the application and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the machine learning-based assisted memory method according to an embodiment of this application; Figure 2 This is a schematic diagram of a machine learning-based assisted memory system according to an embodiment of this application. Detailed Implementation

[0019] Exemplary embodiments of the present application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this application will be thorough and complete, and will fully convey the scope of the present application to those skilled in the art.

[0020] To address the problems existing in the prior art, embodiments of this application provide a machine learning-based assisted memory method and system.

[0021] This application provides a machine learning-based assisted memory method, the process of which is as follows: Figure 1 As shown, it includes the following steps: Step 1: Based on the letter sequence length of the target word, set the length of the user input field for the target word in the user interaction module; obtain the user's input feedback text on the target word through the user interaction module. For example, if the letter sequence of the target word "paper" is ['p', 'a', 'p', 'e', ​​'r'], and its length is 5, then the length of the user input field for the target word in the user interaction module is set to 5. User input that does not meet this length will not be collected, and the user will be reminded to input the word according to this length. This user interaction module can be a tablet computer, mobile smart communication device, smart screen, etc., using the method described in this embodiment.

[0022] Step 2: Obtain the memory feedback letter sequence based on the memory feedback text; calculate the Hamming distance between the letter sequence of the target memory word and the memory feedback letter sequence. If the Hamming distance is greater than 0, mark the corresponding position of the letter sequence of the target memory word as a forgotten position in the Hamming distance; establish word-assisted memory samples based on the letter sequence of the target memory word, its forgotten positions, and the input time of the corresponding memory feedback text; using multiple word-assisted memory samples, with the number of forgotten positions of the letter sequence of the same target memory word decreasing over time as a constraint, set an objective function, and train the machine learning model for display frequency increase / decrease classification to obtain the display frequency increase / decrease classification model.

[0023] If the Hamming distance between the letter sequence of the target word and the letter sequence of the memory feedback is 0, it means that the letters in the same position of the letter sequence of the target word and the letter sequence of the memory feedback are the same. In other words, the letter sequence of the target word and the letter sequence of the memory feedback are completely identical. Therefore, the Hamming distance is 0, indicating that the user's memory of the word is completely correct and there are no forgotten positions. The word can be reviewed according to the review frequency determined based on the existing forgetting curve, or the review frequency can be appropriately reduced.

[0024] Specifically, calculating the Hamming distance between the letter sequence of the target word to be memorized and the letter sequence of the memory feedback includes the following steps: unifying the letters in the letter sequence of the target word to be memorized and the letter sequence of the memory feedback to be lowercase or uppercase; performing an XOR operation on each letter in the letter sequence of the target word to be memorized and the letter sequence of the memory feedback after unifying them to be lowercase or uppercase, and accumulating the results of the XOR operation to obtain the Hamming distance between the letter sequence of the target word to be memorized and the letter sequence of the memory feedback.

[0025] If the Hamming distance is greater than 0, the letter sequence of the target word is included in the Hamming distance and marked as the forgotten position. For example, the letter sequence of the target word "paper" is ['p', 'a', 'p', 'e', ​​'r']. If the user interaction module returns the user input memory feedback letter sequence as ['p', 'a', 'p', 'o', 'r'], then the two letter sequences differ in only one corresponding letter. The letters in the same position are XORed, and the cumulative XOR result is 1. Therefore, the Hamming distance between these two letter sequences is 1, and the corresponding position of the target word's letter sequence in the Hamming distance is the fourth position of that letter sequence, which is the position of the letter 'e'.

[0026] Based on the letter sequence of the target word, its forgotten positions, and the input time of the corresponding memory feedback text, a word-assisted memory sample is established. Using multiple word-assisted memory samples, with the constraint that the number of forgotten positions in the letter sequence of the same target word decreases over time, an objective function is set, and the machine learning model is trained to classify the increase and decrease in display frequency, resulting in a display frequency increase and decrease classification model. For example, for the letter sequence of the target word ['p', 'a', 'p', 'e', ​​'r'], multiple input times and corresponding forgotten positions are collected. To simplify data processing, a mapping table is established for the letter sequences of different target words, establishing a one-to-one correspondence between ['p', 'a', 'p', 'e', ​​'r'] and the word code 0001, and a one-to-one correspondence between the letter sequences of other target words and other word codes. Similarly, the input time is converted into a computationally pleasing encoding form, and the forgotten position code is obtained based on the forgotten position order. If there are no forgotten positions, the forgotten position code can be set to a fixed code such as 1111. Based on the above encoding, a word aid memory sample can be represented as a three-dimensional vector (Word, B(Word, T), T(Word)), where Word represents the letter sequence encoding of the target word, B(Word, T) represents the forgetting position encoding of Word at input time T, and T(Word) represents the input time encoding of Word. A binary classification machine learning model is selected, and multiple word aid memory samples are input into the model. The objective function is set with the constraint that the number of forgetting positions of the same target word decreases over time. The selected binary classification machine learning model is then trained to obtain a display frequency increase / decrease classification model. New vocabulary aids are input into a frequency-based classification model. This model categorizes the input samples into two classes: the first class shows a decrease in the number of forgotten positions in the letter sequence of the target word over time, while the second class shows an increase or no change in the number of forgotten positions. Consequently, the display frequency of words in the second class increases, while the display frequency of words in the first class can be adjusted according to the existing forgetting curve, or the review frequency can be appropriately reduced (as reflected in the display frequency of the word in the user interaction module). In the user interaction module, the review frequency corresponds to the display frequency of the word.

[0027] Based on this frequency increase / decrease classification model, the frequency of word display is adjusted. Compared to memorizing words according to the existing forgetting curve, this method can filter out words that are more easily forgotten and less easily forgotten for the user. Combined with word forgetting positions, it helps users filter out words that are more easily forgotten and specific letter positions that are easily forgotten, allowing users to focus more attention on memorizing the easily forgotten parts. Existing forgetting curve memorization methods cannot filter out easily forgotten and less easily forgotten words from different words memorized at the same time; they can only set up repetitive memorization plans for words memorized at different times. The method described in this embodiment can filter out easily forgotten and less easily forgotten words from different words memorized at the same time and determine different repetitive memorization strategies for them, which helps improve memorization efficiency.

[0028] As a supplement, the machine learning-based assisted memory method further includes the following steps: determining the forgetting rate of the target word based on the Hamming distance between the letter sequence of the target word and the letter sequence of the memory feedback; and determining the display frequency of the target word based on its forgetting rate. For example, a large Hamming distance indicates a large number of forgotten positions; the forgetting rate of the target word is set according to the Hamming distance value, or according to the proportion of forgotten positions to the number of letters in the word. Based on the existing forgetting curve, the estimated review frequency of the word is determined; this estimated review frequency is then adjusted by increasing or decreasing the estimated review frequency based on the forgetting rate of the word to determine the display frequency of the target word. A large forgetting rate corresponds to an increase, and a small forgetting rate corresponds to a decrease. The amount of adjustment can be a fixed value, which can be determined through multiple trials to find a suitable value, or determined based on a linear regression relationship between historical forgetting rates and the amount of change.

[0029] As a supplement, the machine learning-based assisted memorization method further includes the following step: if the number of words used for assisted memorization does not exceed one thousand, then the machine learning model adopts a Support Vector Machine (SVM) model. Since the SVM model performs well in classifying small-scale samples of less than 1000 words, and the number of words used for repeated memorization is usually less than one thousand, using the SVM model can achieve good classification results.

[0030] Step 3: When the target word is displayed in the user interaction module, the display frequency of the target word is updated based on the display frequency increase / decrease classification model, and the letters of the forgotten position of the target word are displayed in a differentiated manner.

[0031] Before or after displaying the target word in the user interaction module, the display frequency of the target word needs to be recalculated. The auxiliary memory sample of the target word is input into a display frequency increase / decrease classification model. This model divides the input samples into two categories: in the first category, the number of forgotten positions in the letter sequence of the target word decreases over time; in the second category, the number of forgotten positions in the letter sequence of the target word increases or remains constant over time. Correspondingly, the display frequency of the second category increases, while the display frequency of the first category can be adjusted according to the existing forgetting curve, or the review frequency can be appropriately reduced. The specific change in display frequency or review frequency can be a fixed value, which can be determined through multiple trials to find a suitable value.

[0032] As a supplement, the letters in the forgotten positions of the target word are displayed in a differentiated manner, including the following steps: highlighting or enlarging the letters in the forgotten positions of the target word, and displaying the letters other than those in the forgotten positions normally. The method of distinguishing the letters in the forgotten positions is not limited to highlighting or enlarging; other display methods such as font outlines, bolding, and different fonts can also be used to attract the user's attention.

[0033] As a supplement, highlighting the letters at the forgotten positions of the target memorized words includes the following steps: Based on the input time of the memory feedback text corresponding to the forgotten position of the target memorized word, the letters at the forgotten positions are highlighted or enlarged in a tiered manner. For example, the longer the input time of the memory feedback text corresponding to the forgotten position of the target memorized word, the worse the memorization effect, and the more attention needs to be attracted. Therefore, forgotten positions with longer input times are displayed more prominently than those with more recent input times, using brighter or larger fonts to attract more user attention and help users focus their attention on the forgotten positions.

[0034] The method described in this embodiment can more finely differentiate the word memorization effect. By combining the forgetting points, it adopts differentiated repetition memorization strategies for words with different degrees of forgetting and distinguishes the letters in the forgotten points of the words, helping users to focus their attention on the more easily forgotten parts, which helps to improve the auxiliary memorization effect.

[0035] Those skilled in the art can change the above-described order of recording or implementation without departing from the scope of protection of this application.

[0036] Another embodiment of this application provides a machine learning-based assisted memory system, such as... Figure 2 As shown, it includes: The user interaction module is configured to: set the length of the user input position of the target word in the user interaction module according to the length of the letter sequence of the target word; obtain the user input feedback text on the target word; and when the target word is displayed in the user interaction module, update the display frequency of the target word based on the display frequency increase / decrease classification model, and differentiate the letters of the forgotten position of the target word. The data processing module is configured to: obtain a memory feedback letter sequence based on the memory feedback text; calculate the Hamming distance between the letter sequence of the target memory word and the memory feedback letter sequence; if the Hamming distance is greater than 0, mark the corresponding position of the letter sequence of the target memory word as a forgotten position in the Hamming distance; establish a word auxiliary memory sample based on the letter sequence of the target memory word, its forgotten positions, and the input time of the corresponding memory feedback text; use multiple word auxiliary memory samples, with the number of forgotten positions of the letter sequence of the same target memory word decreasing over time as a constraint, set an objective function, and train the machine learning model to perform display frequency increase / decrease classification training to obtain a display frequency increase / decrease classification model.

[0037] As a supplement, the data processing module is also configured to: determine the forgetting rate of the target word based on the Hamming distance between the letter sequence of the target word and the letter sequence of the memory feedback; and determine the display frequency of the target word based on the forgetting rate of the target word.

[0038] As a supplement, the data processing module is specifically used to: unify the letters in the target memorized word letter sequence and the memory feedback letter sequence to lowercase or uppercase; perform XOR calculation on each letter in the target memorized word letter sequence and the memory feedback letter sequence after unifying them to lowercase or uppercase, and accumulate the XOR calculation results to obtain the Hamming distance between the target memorized word letter sequence and the memory feedback letter sequence.

[0039] As a supplement, the machine learning model is a support vector machine model.

[0040] Regarding the system in the above embodiments, the specific ways in which each module performs operations have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0041] In this embodiment, the effectiveness of word memorization is further differentiated. By combining the forgetting points, differentiated repetition memorization strategies are adopted for words with different degrees of forgetting. The letters in the forgotten points of a word are displayed in a differentiated manner to help users focus their attention on the more easily forgotten parts, which helps to improve the effect of assisted memorization.

Claims

1. A machine learning-based memory-aided method, characterized in that, Includes the following steps: Based on the letter sequence length of the target word, set the length of the user input field for the target word in the user interaction module; The user interaction module obtains the user's input feedback text on the target word. Based on the memory feedback text, the memory feedback letter sequence is obtained; Calculate the Hamming distance between the letter sequence of the target memorized word and the letter sequence of the memory feedback. If the Hamming distance is greater than 0, then the letter sequence of the target memorized word is included in the corresponding position of the Hamming distance and marked as the forgetting position. When the target word is displayed in the user interaction module, the letters of the forgotten position of the target word are displayed in a differentiated manner.

2. The machine learning-based assisted memory method as described in claim 1, characterized in that, The machine learning-based memory-assisted method further includes the following steps: Based on the letter sequence of the target word, its forgotten positions, and the input time of the corresponding memory feedback text, a word-assisted memory sample is established. Using multiple word-assisted memory samples, with the number of forgotten positions of the letter sequence of the same target word decreasing over time as a constraint, an objective function is set, and the machine learning model is trained to classify the increase and decrease of display frequency, thus obtaining the display frequency increase and decrease classification model. When the target word is displayed in the user interaction module, the display frequency of the target word is updated based on the display frequency increase / decrease classification model.

3. The machine learning-based memory-aided method as described in claim 1, characterized in that, The machine learning-based memory-assisted method further includes the following steps: The forgetting rate of the target word is determined based on the Hamming distance between the letter sequence of the target word and the letter sequence of the memory feedback. The frequency of display of the target words is determined based on their forgetting rate.

4. The machine learning-based memory-aided method as described in claim 1, characterized in that, Calculating the Hamming distance between the letter sequence of the target memorized word and the letter sequence of the memory feedback includes the following steps: The letters in the target word sequence and the memory feedback sequence are unified to lowercase or uppercase. For the target word sequence and the memory feedback sequence after being unified to lowercase or uppercase, XOR calculation is performed on the letters with the same position, and the results of the XOR calculation are accumulated to obtain the Hamming distance between the target word sequence and the memory feedback sequence.

5. The machine learning-based memory-aided method as described in claim 1, characterized in that, The machine learning-based memory-assisted method further includes the following steps: If the number of word-aided memory samples does not exceed one thousand, then the machine learning model adopts the support vector machine model.

6. The machine learning-based memory-aided method as described in claim 1, characterized in that, Distinguishing the forgotten letters of the target word for memorization includes the following steps: Highlight or enlarge the letters in the forgotten positions of the target word.

7. The machine learning-based memory-aided method as described in claim 6, characterized in that, Highlighting the forgotten letters of the target word includes the following steps: Based on the input time of the memory feedback text corresponding to the forgotten position of the target word, the letters of the forgotten position of the target word are highlighted or enlarged in a graded manner.

8. A machine learning-based assisted memory system, characterized in that, include: The user interaction module is configured to: set the length of the user input field for the target word to be memorized based on the length of the letter sequence of the target word; and obtain the user's input feedback text on memorizing the target word. When the target word is displayed in the user interaction module, the display frequency of the target word is updated based on the display frequency increase / decrease classification model, and the letters of the forgotten position of the target word are displayed in a differentiated manner. The data processing module is configured to: obtain a memory feedback letter sequence based on the memory feedback text; calculate the Hamming distance between the letter sequence of the target memory word and the memory feedback letter sequence; if the Hamming distance is greater than 0, mark the corresponding position of the letter sequence of the target memory word as a forgotten position in the Hamming distance; establish a word auxiliary memory sample based on the letter sequence of the target memory word, its forgotten positions, and the input time of the corresponding memory feedback text; use multiple word auxiliary memory samples, with the number of forgotten positions of the letter sequence of the same target memory word decreasing over time as a constraint, set an objective function, and train the machine learning model to perform display frequency increase / decrease classification training to obtain a display frequency increase / decrease classification model.

9. The machine learning-based assisted memory system as described in claim 8, characterized in that, The data processing module is further configured to: determine the forgetting rate of the target word based on the Hamming distance between the letter sequence of the target word and the letter sequence of the memory feedback; and determine the display frequency of the target word based on the forgetting rate of the target word.

10. The machine learning-based assisted memory system as described in claim 8, characterized in that, The data processing module is specifically used to: unify the letters in the target word letter sequence and the memory feedback letter sequence to lowercase or uppercase; perform XOR calculation on each letter in the target word letter sequence and the memory feedback letter sequence after unification to lowercase or uppercase, and accumulate the XOR calculation results to obtain the Hamming distance between the target word letter sequence and the memory feedback letter sequence.

Images