Ranking of segments of learning materials

1. A method comprising:

locating a plurality of occurrences of a knowledge point in a learning material;

selecting one or more candidate initial points from the plurality of occurrences of the knowledge point;

setting a first candidate initial point of the one or more candidate initial points as a first initial point;

creating a first window in the learning material that includes the first initial point, wherein the first window includes a first-window size that corresponds to one or more basic units;

creating a second window in the learning material, wherein a start of the second window follows a start of the first window by a first separation size, and wherein the second window includes a second-window size that corresponds to the one or more basic units;

calculating a window similarity between first-window content of the first window and second-window content of the second window;

in response to the window similarity between the first-window content of the first window and the second-window content of the second window meeting a similarity threshold, generating a first segment with first-segment content that includes at least the first-window content and the second-window content;

detecting a position for a first segment border of the first segment that indicates an end of the first segment in which the detecting includes:

sliding the first window and the second window through the learning material by a step size to create a first new window and a second new window such that the second-window content of the second window is the same as first new-window content of the first new window; and

determining whether a new-window similarity between the first new-window content and second new-window content meet the similarity threshold;

calculating a first-segment consistency measurement for the first segment based on a first-segment similarity between the first-segment content in the first segment and the knowledge point;

ranking, according to one or more computer-executable expressions, the first segment with respect to one or more of the following: a second segment in the learning material and a third segment in a different learning material, wherein the ranking of the first segment is based on one or more of the following: a quality measurement, a learning material type of the learning material, a length of the first segment, and the first-segment consistency measurement of the first segment; and

recommending the first segment to a learner based on the ranking of the first segment.

2. The method of claim 1 , further comprising, in response to the window similarity between the first-window content of the first window and the second-window content of the second window meeting a similarity threshold:

creating the first new window in the learning material, wherein a start of the first new window follows a start of the first window by the step size;

creating the second new window in the learning material, wherein a start of the second new window follows a start of the first new window by a second separation size, and wherein the second new window follows the first window in the learning material;

calculating the new-window similarity between first new-window content of the first new window and second new-window content of the second new window; and

in response to the new-window similarity between the first new-window content of the first new window and the second new-window content of the second new window meeting the similarity threshold, adding the first new-window content and the second new-window content to the first segment; or

in response to the new-window similarity between the first new-window content of the first new window and the second new-window content of the second new window failing to meet the similarity threshold, generating the first segment border at the first new window.

3. The method of claim 1 , wherein selecting the one or more candidate initial points from the plurality of occurrences of the knowledge point includes selecting, as a candidate initial point, an occurrence of the knowledge point located in: a title of a video included in the learning material, a transcript of the video, a title of an article included in the learning material, or a title corresponding to one or more slides included in the learning material.

4. The method of claim 1 , further comprising:

setting a score for each of the plurality of occurrences of the knowledge point in the learning material, wherein the score for an occurrence of the knowledge point in a title of the learning material is higher than the score for an occurrence of the knowledge point in a text of the learning material; and

adjusting the first-segment consistency measurement based on a sum of the scores for each of the plurality of occurrences of the knowledge point in the learning material.

5. The method of claim 1 , further comprising:

creating a third new window in the learning material;

creating a fourth new window in the learning material that is adjacent to the third new window;

calculating a second window similarity between third new-window content of the third new window and fourth new-window content of the fourth new window;

in response to the second window similarity between the third new-window content of the third new window and the fourth new-window content of the fourth new window meeting a similarity threshold, generating the second segment that includes at least the third new-window content and the fourth new-window content;

calculating a second-segment consistency measurement for the second segment, wherein the second-segment consistency measurement for the second segment is based on a similarity between a second-segment content in the second segment and the knowledge point; and

ranking the second segment with respect to the first segment in the learning material based on one or more of the following: a quality measurement, the learning material type of the learning material, a length of the second segment, and the second-segment consistency measurement of the second segment.

6. The method of claim 1 , wherein:

the first separation size corresponds to one or more basic units;

the first-window size corresponds to a same number of the basic unit as the second-window size;

the first window and the second window overlap in response to the first-window size corresponding to a greater number of the basic unit than the first separation size; and

the first window is adjacent to the second window in response to the first-window size corresponding to a same number of the basic unit as the first separation size.

7. The method of claim 1 , wherein:

in response to the learning material type of the learning material including a video, the one or more basic units include one or more sentences of a transcript for the video;

in response to the learning material type of the learning material including an article, the one or more basic units include one or more sentences of the article; and

in response to the learning material type of the learning material including one or more slides, the one or more basic units include a quantity of slides.

8. The method of claim 1 , wherein:

in response to the first segment being located in a video, the length of the first segment is based on a duration of the first segment;

in response to first segment being located in an article or text, the length of the first segment is based on a number of words or bytes in the first segment; and

in response to the first segment being located in one or more slides, the length of the first segment is based on a quantity of slides in the first segment.

9. The method of claim 1 , wherein the first segment matches a learner preference that includes one or more of the following: a learning material type, a window size, a learning-specific feature, a learning material age, a segment age, and a segment length.

10. A system, comprising:

one or more processors;

non-transitory computer readable media that include instructions thereon that, in response to execution by the one or more processors, control performance of operations comprising:

locating a plurality of occurrences of a knowledge point in a learning material;

selecting one or more candidate initial points from the plurality of occurrences of the knowledge point;

setting a first candidate initial point of the one or more candidate initial points as a first initial point;

creating a first window in the learning material that includes the first initial point, wherein the first window includes a first-window size that corresponds to one or more basic units;

creating a second window in the learning material that is adjacent to the first window, wherein a start of the second window follows a start of the first window by a first separation size, and wherein the second window includes a second-window size that corresponds to the one or more basic units;

calculating a window similarity between first-window content of the first window and second-window content of the second window;

in response to the window similarity between the first-window content of the first window and the second-window content of the second window meeting a similarity threshold, generating a first segment with first-segment content that includes at least the first-window content and the second-window content;

detecting a position for a first segment border of the first segment that indicates an end of the first segment in which the detecting includes:

sliding the first window and the second window through the learning material by a step size to create a first new window and a second new window such that the second-window content of the second window is the same as first new-window content of the first new window, and

determining whether a new-window similarity between the first new-window content and second new-window content meet the similarity threshold;

calculating a first-segment consistency measurement for the first segment based on a first-segment similarity between the first-segment content in the first segment and the knowledge point;

ranking, according to one or more computer-executable expressions, the first segment with respect to one or more of the following: a second segment in the learning material and a third segment in a different learning material, wherein the ranking of the first segment is based on one or more of the following: a quality measurement, a learning material type of the learning material, a length of the first segment, and the first-segment consistency measurement of the first segment; and

recommending the first segment to a learner based on the ranking of the first segment.

11. The system of claim 10 , wherein in response to the window similarity between the first-window content of the first window and the second-window content of the second window meeting a similarity threshold, an analysis server is configured to generate the first segment with the first-segment content that includes at least the first-window content and the second-window content by being configured to:

create the first new window in the learning material, wherein a start of the first new window follows a start of the first window by the step size;

create the second new window in the learning material, wherein a start of the second new window follows a start of the first new window by a second separation size, and wherein the second new window follows the first window in the learning material;

calculate the new-window similarity between first new-window content of the first new window and second new-window content of the second new window; and

in response to the new-window similarity between the first new-window content of the first new window and the second new-window content of the second new window meeting the similarity threshold, add the first new-window content and the second new-window content to the first segment; or

in response to the new-window similarity between the first new-window content of the first new window and the second new-window content of the second new window failing to meet the similarity threshold, generate the first segment border at the first new window.

12. The system of claim 10 , wherein an analysis server is configured to select one or more candidate initial points from the plurality of occurrences of the knowledge point by being configured to select, as a candidate initial point, an occurrence of the knowledge point located in: a title of a video included in the learning material, a transcript of the video, a title of an article included in the learning material, or a title corresponding to one or more slides included in the learning material.

13. The system of claim 10 , wherein an analysis server is configured to:

set a score for each of the plurality of occurrences of the knowledge point in the learning material, wherein the score for an occurrence of the knowledge point in a title of the learning material is higher than the score for an occurrence of the knowledge point in a text of the learning material; and

adjust the first-segment consistency measurement based on a sum of the scores for each of the plurality of occurrences of the knowledge point in the learning material.

14. The system of claim 10 , wherein an analysis server is configured to:

create a third new window in the learning material;

create a fourth new window in the learning material that is adjacent to the third new window;

calculate a second window similarity between third new-window content of the third new window and fourth new-window content of the fourth new window;

in response to the second window similarity between the third new-window content of the third new window and the fourth new-window content of the fourth new window meeting a similarity threshold, generate the second segment that includes at least the third new-window content and the fourth new-window content;

calculate a second-segment consistency measurement for the second segment, wherein the second-segment consistency measurement for the second segment is based on a similarity between a second-segment content in the second segment and the knowledge point; and

rank the second segment with respect to the first segment in the learning material based on one or more of the following: a quality measurement, the learning material type of the learning material, a length of the second segment, and the second-segment consistency measurement of the second segment.

15. The system of claim 10 , wherein the first segment matches a learner preference that includes one or more of the following: a learning material type, a window size, a learning-specific feature, a learning material age, a segment age, and a segment length.

16. A non-transitory computer-readable medium that includes computer-readable instructions stored thereon that are executable by a processor to perform or control performance of operations comprising:

locating a plurality of occurrences of a knowledge point in a learning material;

selecting one or more candidate initial points from the plurality of occurrences of the knowledge point;

setting a first candidate initial point of the one or more candidate initial points as a first initial point;

creating a first window in the learning material that includes the first initial point, wherein the first window includes a first-window size that corresponds to one or more basic units;

creating a second window in the learning material, wherein a start of the second window follows a start of the first window by a first separation size, and wherein the second window includes a second-window size that corresponds to the one or more basic units;

calculating a window similarity between first-window content of the first window and second-window content of the second window;

in response to the window similarity between the first-window content of the first window and the second-window content of the second window meeting a similarity threshold, generating a first segment with first-segment content that includes at least the first-window content and the second-window content;

detecting a position for a first segment border of the first segment that indicates an end of the first segment in which the detecting includes:

sliding the first window and the second window through the learning material by a step size to create a first new window and a second new window such that the second-window content of the second window is the same as a first new-window content of the first new window; and

determining whether a new-window similarity between the first new-window content and a second new-window content meet the similarity threshold;

calculating a first-segment consistency measurement for the first segment based on a first-segment similarity between the first-segment content in the first segment and the knowledge point;

ranking, according to one or more computer-executable expressions, the first segment with respect to one or more of the following: a second segment in the learning material and a third segment in a different learning material, wherein the ranking of the first segment is based on one or more of the following: a quality measurement, a learning material type of the learning material, a length of the first segment, and the first-segment consistency measurement of the first segment; and

recommending the first segment to a learner based on the ranking of the first segment.

17. The non-transitory computer-readable medium of claim 16 , wherein the operations further comprise, in response to the window similarity between the first-window content of the first window and the second-window content of the second window meeting a similarity threshold:

creating the first new window in the learning material, wherein a start of the first new window follows a start of the first window by the step size;

creating the second new window in the learning material, wherein a start of the second new window follows a start of the first new window by a second separation size, and wherein the second new window follows the first window in the learning material;

calculating the new-window similarity between first new-window content of the first new window and second new-window content of the second new window; and

in response to the new-window similarity between the first new-window content of the first new window and the second new-window content of the second new window meeting the similarity threshold, adding the first new-window content and the second new-window content to the first segment; or

in response to the new-window similarity between the first new-window content of the first new window and the second new-window content of the second new window failing to meet the similarity threshold, generating the first segment border at the first new window.

18. The non-transitory computer-readable medium of claim 16 , wherein selecting the one or more candidate initial points from the plurality of occurrences of the knowledge point includes selecting, as a candidate initial point, an occurrence of the knowledge point located in: a title of a video included in the learning material, a transcript of the video, a title of an article included in the learning material, or a title corresponding to one or more slides included in the learning material.

19. The non-transitory computer-readable medium of claim 16 , wherein the first segment matches a learner preference that includes one or more of the following: a learning material type, a window size, a learning-specific feature, a learning material age, a segment age, and a segment length.

20. The method of claim 1 , wherein the one or more computer-executable expressions include:

Consistency Measurement=Similarity( {right arrow over (q)}, {right arrow over (e)} )=cos(θ), 0<cos(θ) <1;

a×QM+b×CM+c×L+d×TM; and

a+b+c+d= 1,

wherein:

{right arrow over (q)} represents a term vector of a knowledge point or a query;

{right arrow over (e)} represents a term vector of a segment;

cos(θ) represents the cosine of an angle θ between the term vectors {right arrow over (q)} and {right arrow over (e)};

QM represents a quality measurement for a segment;

CM represents a consistency measurement for a segment;

L represents a length of a segment;

TM represents a learning material type;

a, b, c, and d represent weights corresponding to the quality measurement, the consistency measurement, the length of the segment, and the learning material type, respectively; and

× is a scalar multiplier.