Method and system for content bias detection

1. A method, implemented on a machine having at least one processor, storage, and a communication platform capable of connecting to a network for providing content, the method comprising:

obtaining a plurality of content items and publication information associated with each of the plurality of content items;

determining one or more topics of each of the plurality of content items;

analyzing the publication information associated with each of the plurality of content items to identify at least one source of a plurality of sources that published the content item;

determining, with respect to each of the plurality of sources, sentiment features of content items published by the source, wherein the sentiment features are determined in accordance with a sentiment feature model derived via machine learning on training data;

generating a source-topic archive storing information indicative of a bias of each of the plurality of sources publishing content based on sentiment features of the published content, wherein the source-topic archive comprises, with respect to each of the one or more topics and each of the plurality of sources:

a first quantity of one or more of the plurality of content items published by the source on the topic with a positive sentiment, and an associated first sentiment indicator indicative of the positive sentiment, and

a second quantity of one or more of the plurality of content items published by the source on the topic with a negative sentiment, and an associated second sentiment indicator indicative of the negative sentiment;

in response to a search query from a user for a query content item of the plurality of content items on a query topic published by a first source of the plurality of sources, identifying, based on the first quantity and the second quantity associated with the query topic of the query content item associated with the first source, a bias of the first source with respect to the query topic; and

providing the user with information indicative of the bias of the first source.

2. The method of claim 1 , wherein the bias of the first source is a coverage bias with respect to the query topic, the coverage bias being determined based on a first number of content items of the query topic published by the first source.

3. The method of claim 2 , wherein the first source is determined to have the coverage bias with respect to the query topic based on the first number of content items of the query topic published by the first source being zero.

4. The method of claim 2 , wherein the first source is determined to have the coverage bias with respect to the query topic based on a difference between the first number of content items of the query topic published by the first source and a second number of content items of the query topic published by other of the plurality of sources being greater than a predetermined threshold.

5. The method of claim 1 , wherein the bias of the first source is a sentiment bias with respect to the query topic, the sentiment bias corresponding to a manner in which content items of the query topic are published by the first source.

6. The method of claim 5 , further comprising:

obtaining the content items associated with the query topic that are published by the first source;

extracting, in accordance with the sentiment feature model, one or more sentiment features from each of the content items; and

determining the sentiment bias based on the extracted one or more sentiment features from the content items.

7. The method of claim 5 , wherein the sentiment bias is one of a positive sentiment, a negative sentiment, and a neutral sentiment.

8. A machine readable and non-transitory medium having information recorded thereon for providing content, wherein the information, when read by the machine, causes the machine to perform:

obtaining a plurality of content items and publication information associated with each of the plurality of content items;

determining one or more topics of each of the plurality of content items;

analyzing the publication information associated with each of the plurality of content items to identify at least one source of a plurality of sources that published the content item;

determining, with respect to each of the plurality of sources, sentiment features of content items published by the source, wherein the sentiment features are determined in accordance with a sentiment feature model derived via machine learning on training data;

generating a source-topic archive storing information indicative of a bias of each of the plurality of sources publishing content based on sentiment features of the published content, wherein the source-topic archive comprises, with respect to each of the one or more topics and each of the plurality of sources:

a first quantity of one or more of the plurality of content items published by the source on the topic with a positive sentiment, and an associated first sentiment indicator indicative of the positive sentiment, and

a second quantity of one or more of the plurality of content items published by the source on the topic with a negative sentiment, and an associated second sentiment indicator indicative of the negative sentiment;

in response to a search query from a user for a query content item of the plurality of content items on a query topic published by a first source of the plurality of sources, identifying, based on the first quantity and the second quantity associated with the query topic of the query content item associated with the first source, a bias of the first source with respect to the query topic; and

providing the user with information indicative of the bias of the first source.

9. The medium of claim 8 , wherein the bias of the first source is a coverage bias with respect to the query topic, the coverage bias being determined based on a first number of content items of the query topic published by the first source.

10. The medium of claim 9 , wherein the first source is determined to have the coverage bias with respect to the query topic based on the first number of content items of the query topic published by the first source being zero.

11. The medium of claim 9 , wherein the first source is determined to have the coverage bias with respect to the query topic based on a difference between the first number of content items of the query topic published by the first source and a second number of content items of the query topic published by other of the plurality of sources being greater than a predetermined threshold.

12. The medium of claim 8 , wherein the bias of the first source is a sentiment bias with respect to the query topic, the sentiment bias corresponding to a manner in which content items of the query topic are published by the first source.

13. The medium of claim 12 , wherein the information, when read by the machine, further causes the machine to perform:

obtaining the content items associated with the query topic that are published by the first source;

extracting, in accordance with the sentiment feature model, one or more sentiment features from each of the content items; and

determining the sentiment bias based on the extracted one or more sentiment features from the content items.

14. The medium of claim 12 , wherein the sentiment bias is one of a positive sentiment, a negative sentiment, and a neutral sentiment.

15. A system for providing content comprising:

a content retrieval unit implemented by a processor and configured for obtaining a plurality of content items and publication information associated with each of the plurality of content items;

a topic determining unit implemented by a processor and configured for determining one or more topics of each of the plurality of content items;

a content processing unit implemented by a processor and configured for analyzing the publication information associated with each of the plurality of content items to identify at least one source of a plurality of sources that published the content item;

a sentiment feature extractor implemented by a processor and configured for determining, with respect to each of the plurality of sources, sentiment features of content items published by the source, wherein the sentiment features are determined in accordance with a sentiment feature model derived via machine learning on training data;

a clustering unit implemented by a processor and configured for generating a source-topic archive storing information indicative of a bias of each of the plurality of sources publishing content based on sentiment features of the published content, wherein the source-topic archive comprises, with respect to each of the one or more topics and each of the plurality of sources:

a first quantity of one or more of the plurality of content items published by the source on the topic with a positive sentiment, and an associated first sentiment indicator indicative of the positive sentiment, and

a second quantity of one or more of the plurality of content items published by the source on the topic with a negative sentiment, and an associated second sentiment indicator indicative of the negative sentiment;

a bias determining unit implemented by a processor and configured for in response to a search query from a user for a query content item of the plurality of content items on a query topic published by a first source of the plurality of sources, identifying, based on the first quantity and the second quantity associated with the query topic of the query content item associated with the first source, a bias of the first source with respect to the query topic; and

a search engine implemented by a processor and configured for providing the user with information indicative of the bias of the first source.

16. The system of claim 15 , wherein the bias of the first source is a coverage bias with respect to the query-a topic, the coverage bias being determined based on a first number of content items of the query topic published by the first source.

17. The system of claim 16 , wherein the first source is determined to have the coverage bias with respect to the query topic based on the first number of content items of the query topic published by the first source being zero.

18. The system of claim 16 , wherein the first source is determined to have the coverage bias with respect to the query topic based on a difference between the first number of content items of the query topic published by the first source and a second number of content items of the query topic published by other of the plurality of sources being greater than a predetermined threshold.

19. The system of claim 15 , wherein the bias of the first source is a sentiment bias with respect to the query topic, the sentiment bias corresponding to a manner in which content items of the query topic are published by the first source.

20. The system of claim 19 ,

wherein the content retrieval unit is implemented by a processor and configured for further obtaining the content items associated with the query topic that are published by the first source, and

wherein the sentiment feature extractor is implemented by a processor and configured for extracting, in accordance with the sentiment feature model, one or more sentiment features from each of the content items; and

the system further comprising a sentiment based analyzer implemented by a processor and configured for determining the sentiment bias based on the extracted one or more sentiment features from the content items.

21. The system of claim 19 , wherein the sentiment bias is one of a positive sentiment, a negative sentiment, and a neutral sentiment.