Personalized learning system and method for the automated generation of structured learning assets based on user data
Learning systems and methods of the present disclosure include generating a text document based on a digital file, tokenizing the text document, generating a semantic model based on the tokenized text document using an unsupervised machine learning algorithm, assigning a plurality of passage scores to a corresponding plurality of passages of the tokenized text document, selecting one or more candidate knowledge items from the tokenized text document based on the plurality of passage scores, filtering the one or more candidate knowledge items based on user data, generating one or more structured learning assets based on the one or more filtered candidate knowledge items, generating an interaction based at least on the one or more structured learning assets, and transmitting the interaction to a user device. Each passage score is assigned based on a relationship between a corresponding passage and the semantic model.
1. A learning system comprising:
a non-transitory memory; and
one or more hardware processors configured to read instructions from the non-transitory memory to cause the system to perform operations comprising:
generating a text document based on a digital file;
tokenizing the text document;
generating a semantic model based on the tokenized text document using an unsupervised machine learning algorithm;
assigning a plurality of passage scores to a corresponding plurality of passages of the tokenized text document, wherein each passage score is assigned based on a relationship between a corresponding passage and the semantic model;
selecting one or more candidate knowledge items from the tokenized text document based on the plurality of passage scores;
filtering the one or more candidate knowledge items based on user data;
generating one or more structured learning assets based on the one or more filtered candidate knowledge items;
generating an interaction based at least on the one or more structured learning assets; and
transmitting the interaction to a user device.
2. The learning system of claim 1 , wherein the user data comprises direct user feedback.
3. The learning system of claim 1 , wherein the interaction includes a voice interaction.
4. The learning system of claim 3 , wherein the operations further comprise:
receiving a voice response from the user device in reply to the voice interaction;
extracting a user answer from the voice response; and
determining whether the user answer corresponds with a correct answer to the voice interaction.
5. The learning system of claim 4 , wherein the operations further comprise:
generating a multimedia interaction based on the voice response; and
transmitting the multimedia interaction to the user device to output the multimedia interaction.
6. The learning system of claim 1 , wherein generating the text document further comprises:
generating a plurality of intermediate text documents from the digital file using a corresponding plurality of conversion techniques; and
selecting the text document from the plurality of intermediate text documents.
7. The learning system of claim 6 , wherein generating the text document further comprises repairing the text document in response to selecting the text document from the plurality of intermediate text documents.
8. The learning system of claim 1 , wherein the one or more candidate knowledge items are selected adaptively to reduce redundancies among the one or more candidate knowledge items.
9. The learning system of claim 8 , wherein adaptively selecting the one or more candidate knowledge items comprises reducing a passage score when a corresponding passage includes a concept covered in another passage previously selected as one of the one or more candidate knowledge items.
10. The learning system of claim 1 , wherein the operations further comprise:
identifying a plurality of concepts in the tokenized text document; and
assigning a concept score to each of the plurality of concepts based on a second relationship between each of the plurality of concepts and the semantic model;
wherein the plurality of passage scores are assigned based on the concept score corresponding to each of the plurality of concepts.
11. The learning system of claim 10 , wherein the plurality of concepts include noun phrases determined using natural language processing of the tokenized text document, wherein the natural language processing includes at least part of speech tagging and noun chunking.
12. The learning system of claim 10 , wherein the second relationship is defined by a centrality metric that indicates how central a corresponding concept is to one or more topics in the semantic model.
13. The learning system of claim 10 , wherein the concept score is further assigned based on a specificity of the concept.
14. The learning system of claim 1 , wherein generating the text document comprises retrieving metadata associated with the digital file from an external resource.
15. A method comprising:
extracting text from digital content;
performing semantic analysis of the text to generate a semantic model;
scoring each passage of the text based on a relationship between the passage and the semantic model;
selecting one or more candidate knowledge items from the text based on the scoring;
filtering the one or more candidate knowledge items based on user data; and
generating a structured learning asset based on the one or more filtered candidate knowledge items.
16. The method of claim 15 , wherein selecting the one or more candidate knowledge items includes adaptively selecting the one or more candidate knowledge items to reduce redundancies among the one or more candidate knowledge items.
17. The method of claim 15 , further comprising:
scoring each concept of the text based on a relationship between the concept and the semantic model;
wherein the scoring of each passage is based on the scoring of each concept included in the passage.
18. The method of claim 15 , further comprising:
generating an interaction based on the structured learning asset; and
transmitting the interaction to a user device.
19. A non-transitory machine-readable medium having stored thereon machine-readable instructions executable to cause a machine to perform operations comprising:
generating a text document based on received digital content;
tokenizing the text document;
performing semantic analysis of the tokenized text document to generate a semantic model;
scoring each passage of the tokenized text document based on a relationship between the passage and the semantic model;
selecting one or more candidate knowledge items from the tokenized text document based on the scoring;
filtering the one or more candidate knowledge items based on user data; and
generating a structured learning asset based on the one or more filtered candidate knowledge items.
20. The non-transitory machine-readable medium of claim 19 , wherein generating the text document further comprises:
generating a plurality of intermediate text documents from the received digital content using a corresponding plurality of conversion techniques; and
selecting the text document from the plurality of intermediate text documents.