TWI494892B - Recommended Method and System of Internet Learning Community - Google Patents

Description

Network learning community recommendation method and system thereof

The present invention relates to a network learning community recommendation method and system thereof, and more particularly to a network learning community recommendation method and system for collaborative learning.

With the development of the network learning model, the digital learning model has gradually changed from focusing on individual self-learning to a cooperative learning model formed by many people, which has led to the concept of learning community. The learning community is a group of people who expand their collective knowledge and abilities by communicating, sharing knowledge and building common goals to create collective learning behaviors. Through the creation and establishment of the community, the knowledge, resources, information, experience, and experience scattered among different learners can be gathered in the community, so that all learners can share the common, reflect on the growth together, and construct a new individual. A useful body of knowledge.

Learning through the community can not only promote the interaction of learners, but also enhance the learning performance of learners. However, how to assign a new learner to a suitable learning community for learning usually takes a lot of time and effort to assess the suitability of each learning community. This type of pre-assessment is time-consuming and labor-intensive. It is also easy to make an evaluation error due to human error, so it is necessary to seek a solution.

Accordingly, it is an object of the present invention to provide a web-based learning community recommendation method that recommends that the learner to be classified is suitable for each learning community.

Therefore, the network learning community recommendation method of the present invention is applicable to a network learning community recommendation system, and the network learning community recommendation system includes a database, a user interface module, and a k nearest neighbor calculation. The module and a recommendation module, wherein the database stores a plurality of classified learner samples, each of the classified learner samples having a plurality of learning attribute values. The method comprises the following steps: (A) a learner module to use the user interface module to input a learner sample to be classified having a plurality of learning attribute values; (B) the k nearest neighbor calculation module according to each Classifying the learning attribute value of the learner sample and the learning attribute value of the learner sample to be classified, and calculating a predetermined number k of classified learner samples closest to the learner sample to be classified, wherein the k distances are closest The classified learner samples belong to the M learning communities; and (C) the number of the k classified learner samples included in the learning community according to the learning community of the M learning communities Sorting out the learner samples to be classified into a priority order of each learning community to be recommended to the learner to be classified, wherein the recommendation module includes the most of the k classified communities among the M learning communities The learning community of the learner sample is designated as having the highest priority recommendation to the learner to be classified.

Another object of the present invention is to provide a network-based learning community that recommends that the learners to be classified are suitable for each learning community. Recommended system.

Therefore, the network learning community recommendation system of the present invention comprises a database, a user interface module, a k nearest neighbor computing module and a recommendation module. The database is used to store a plurality of classified learner samples, wherein each of the classified learner samples has a plurality of learning attribute values. The user interface module is configured to input a plurality of learning attribute values of the learner sample to be classified. The k nearest neighbor calculation module is configured to calculate, according to the learning attribute value of each classified learner sample and the learning attribute value of the learner sample to be classified, a predetermined number k of distances closest to the to-be-classified learner sample A sample of learners has been classified, wherein the k consecutively classified learner samples belong to M learning communities. The recommendation module is configured to sort the sample of the learner to be classified into each learning community according to the number of the k classified learner samples included in the learning community of each of the M learning communities a priority order is recommended to the learner to be classified, wherein the recommendation module designates the learning communities of the M learning communities that contain the most of the k classified learner samples as having the highest priority recommendation The learner to be classified.

The effect of the present invention is that the k nearest neighbor calculation module calculates k sampled learner samples that are closest to the sample to be classified, and recommends the learner to be classified by the recommendation module. By learning the community's prioritization, it is not necessary to spend time and effort to evaluate the community when learning to learn, and to reduce the risk of miscalculation due to human error.

1‧‧‧Web-based learning community recommendation system

11‧‧‧Database

12‧‧‧User interface module

13‧‧‧k nearest neighbor calculation module

14‧‧‧Recommended module

21~23‧‧‧Steps

Other features and effects of the present invention will be apparent from the following description of the drawings, wherein: FIG. 1 is a system block diagram illustrating a preferred embodiment of the networked learning community recommendation system of the present invention; 2 is a flow chart showing a preferred embodiment of the network learning community recommendation method of the present invention.

Referring to FIG. 1 , a preferred embodiment of the network learning community recommendation system 1 of the present invention includes a database 11 , a user interface module 12 , a k nearest neighbor computing module 13 , and a recommendation module 14 .

The database 11 is used to store a plurality of classified learner samples, wherein each of the classified learner samples has a plurality of learning attribute values.

The user interface module 12 is configured to input a plurality of learning attribute values of the learner sample to be classified.

The k nearest neighbor calculation module 13 is configured to calculate, according to the learning attribute value of each classified learner sample and the learning attribute value of the learner sample to be classified, a predetermined number k of distances from the learner sample to be classified. The classified learner sample, wherein the k consecutively classified learner samples belong to M learning communities.

The recommendation module 14 is configured to sort the sample of the learner to be classified into each learning society according to the number of the k classified learner samples included in each of the M learning communities. A priority order of the group is recommended to the learner to be classified. The recommendation module 14 includes the number of the k classified learner samples in the M learning communities. The zero learning community is treated as an unsuitable learning community and excluded, and only the learning communities that contain the k-classified learner samples that are not zero in the M learning communities are ranked. And the recommendation module 14 designates the learning community including the most of the k classified learner samples among the M learning communities as the highest priority recommendation to the learner to be classified.

Referring to FIG. 1 and FIG. 2, a preferred embodiment of the network learning community recommendation method of the present invention is applicable to the network learning community recommendation system 1, and the method includes the following steps.

First, as shown in step 21, the learner to be classified uses the user interface module 12 to input the sample of the learner to be classified having a plurality of learning attribute values, wherein the learner values are based on the learner to be classified. Obtained by an operation on a digital learning platform system (not shown), wherein the digital learning platform system presents a textbook of each chapter in a webpage manner and can be used by the learner to be uploaded and can record the to-be-categorized The learner's operation record, the establishment of the digital learning platform system is a conventional technology, and therefore will not be described here. The learning attribute values include a time sum of the learner's system using the digital learning platform, the learner's reading chapter of the to-be-classified student, and a time value of one of the chapters, and a time for the learner to read each chapter. The value, the value to be read by the learner to be repeated, and the value of the job that has been uploaded by the learner to be classified account for a percentage of the total job.

Next, as shown in step 22, the k nearest neighbor calculation module 13 learns the learning attribute value of each classified learner sample from the to-be-classified Calculating the learned attribute value of the learner sample, and calculating the predetermined number k of the classified learner samples closest to the learner sample to be classified, wherein the k consecutively classified learner samples belong to the M learning communities .

In the preferred embodiment, the M value is 3, and the M learning communities include a slow learning community, a short-lived learning community, and an in-depth learning community. The learners in the slow learning community spend the longest time reading, and have the most repeated clicks on the behavior of each chapter. The learners in this ephemeral learning community have the shortest reading time, and the behavior of repeating each chapter is the least, and the order of reading the chapters of the course is very irregular. The depth of the course of the learner's browsing in the in-depth learning community is the most in-depth, and the order of the chapters in the course is very regular.

It is worth mentioning that the k-nearest neighbor module 13 is calculated using the following equation is obtained based learning attribute value learning value of the property P _new sample to be classified and the learner of the learner sample i has been classified in the distance d L _{s_i} ( P _new , L _{s_i} ): The learner sample P _new to be classified and each of the classified learner samples L _{s_i} have N learning attribute values, and at _n (P _new ) represents the nth learning attribute value of the learner sample P _new to be classified, at _n (L _{s_i)} represents the i th sample L _{s_i} classified learner learning the attribute values of n, Z for these samples the number of learners of L _{s_i} classified. Furthermore, the k-nearest neighbor module 13 is further calculated based on the distance d (P _new, L _{s_i)} and conventional k-nearest neighbor (of kNN) algorithm, and the like of the k nearest classified learner to classify samples to M Learn the community.

In the preferred embodiment, the predetermined number k has a value of 5, such that (c _i , r _i ) represents the i- th learning community c _i among the k consecutively classified learner samples. The number is r _i , i =1, 2, 3, then r ₁ + r ₂ + r ₃ = 5.

Then, as shown in step 23, the recommendation module 14 sorts the to-category according to the number r _{i of the} k classified learner samples included in each of the M learning communities. The priority of the learner sample belonging to each learning community is recommended to the learner to be classified. Wherein the recommendation module 14 only those M learning community contains such a number k of samples classified learner r _i is not zero learning community sort the recommender module 14 and the other M The learning community that contains the most of the k classified learner samples in the learning community is designated as having the highest priority recommendation to the learner to be classified.

In summary, the k nearest neighbor calculation module 13 calculates the distance between the k and the learner sample to be classified according to the learning attribute value of each classified learner sample and the learning attribute value of the learner sample to be classified. The recently-sorted learner samples are sorted out by the recommendation module 14 to the priority order of the learner group to be classified for each learning community, and recommended to the learner to be selected. At that time, it is not necessary to spend time and effort to carry out the evaluation, and to reduce the risk of misjudging due to human error, it is indeed possible to achieve the object of the present invention.

The above is only the preferred embodiment of the present invention, and the scope of the present invention is not limited thereto, that is, the simple equivalent changes and modifications made by the patent application scope and patent specification content of the present invention, All remain within the scope of the invention patent.

1‧‧‧Web-based learning community recommendation system

11‧‧‧Database

12‧‧‧User interface module

13‧‧‧k nearest neighbor calculation module

14‧‧‧Recommended module

Claims (10)

A network learning community recommendation method is applicable to a network learning community recommendation system, the network learning community recommendation system includes a database, a user interface module, and a k nearest neighbor computing module. And a recommendation module, wherein the database stores a plurality of classified learner samples, each of the classified learner samples has a plurality of learning attribute values, and the method comprises the following steps: (A) a learner to be classified uses the The user interface module inputs a sample of the learner to be classified having a plurality of learning attribute values; (B) the k nearest neighbor calculation module calculates the learning attribute value of each classified learner sample and the sample of the learner to be classified Learning attribute values, calculating a predetermined number k of classified learner samples closest to the sample of the learner to be classified, wherein the k consecutively classified learner samples belong to M learning communities, wherein k≧ 1, M≧2; and (C) the recommendation module sorts the learning to be classified according to the number of the k classified learner samples included in each of the M learning communities Sample A priority order belonging to each learning community is recommended to the learner to be classified, wherein the recommendation module specifies the learning communities of the M learning communities that contain the most of the k classified learner samples as The highest priority is recommended to the learner to be classified. The network learning community recommendation method according to claim 1, wherein in the step (A), the learning attribute values include the learner to be classified The chapter occupies a scale value of one of the chapters, a time value of the chapter to be read by the learner to be classified, and a value of the chapter that the learner to be classified repeatedly reads each chapter. The network learning community recommendation method according to claim 2, wherein in the step (B), the M learning communities comprise a slow learning community, a short learning community, and an in-depth type. Learn the community. The network learning community recommendation method according to claim 1, wherein in the step (B), the k nearest neighbor computing module determines the learning attribute value of the learner sample P _new to be classified by using the following formula: The distance d ( P _new , L _{s_i} ) from the learning attribute value of the i- th classified learner sample L _{s_i} : The learner sample P _new to be classified and each of the classified learner samples L _{s_i} have N learning attribute values, and at _n (P _new ) represents the nth learning attribute value of the learner sample P _new to be classified, at _n (L _{s_i)} represents the i th sample L _{s_i} classified learner learning the attribute values of n, Z for these samples the number of learners of L _{s_i} classified. The network learning community recommendation method according to claim 1, wherein in the step (C), the recommendation module includes the number of the k classified learner samples in the M learning communities. A learning community with zero is considered as an unsuitable learning community and is excluded, and only the learning communities that contain the number of samples of the k classified learners that are not zero are sorted among the M learning communities. . A web-based learning community recommendation system that includes: a database for storing a plurality of classified learner samples, wherein each of the classified learner samples has a plurality of learning attribute values; and a user interface module for inputting a plurality of learning samples of the learner to be classified Attribute value; a k nearest neighbor calculation module, configured to calculate a predetermined number of k and the learner to be classified according to the learning attribute value of each classified learner sample and the learning attribute value of the learner sample to be classified The sample is from the most recent sampled learner sample, wherein the k consecutively classified learner samples belong to M learning communities, where k≧1, M≧2; and a recommendation module for The number of the k classified learner samples included in each learning community of the M learning communities, sorting out a priority order of the learner samples to be classified into each learning community, and recommending to the classified The learner, wherein the recommendation module specifies the learning communities of the M learning communities that contain the most of the k classified learner samples as the highest priority recommendation to the learner to be classified. The network learning community recommendation system according to claim 6, wherein the learning attribute value includes a ratio of the reading chapter of the to-be-classified learner to one of the chapters, and a time for the learner to read each chapter. The value and the learner to be classified repeatedly read the values of each chapter. The web-based learning community recommendation system of claim 7, wherein the M learning communities comprise a slow learning community, a short-lived learning community, and an in-depth learning community. The network learning community recommendation system according to claim 6, wherein the k nearest neighbor computing module uses the following formula to obtain the learning attribute value of the to-be-classified learner sample P _new and the i- th classified learning The distance d ( P _new , L _{s_i} ) of the learning attribute value of the sample L _{s_i} : The learner sample P _new to be classified and each of the classified learner samples L _{s_i} have N learning attribute values, and at _n (P _new ) represents the nth learning attribute value of the learner sample P _new to be classified, at _n (L _{s_i)} represents the i th sample L _{s_i} classified learner learning the attribute values of n, Z for these samples the number of learners of L _{s_i} classified. The network learning community recommendation system according to claim 6, wherein the recommendation module regards the learning communities including the number of the k classified learner samples in the M learning communities as zero An unsuitable learning community is excluded and only the learning communities that contain the k-classified learner samples that are not zero in the learning community are sorted.