System for generating topic-based sentiment time series from social media data
A method for issuing control signals in response to sentiment. In some embodiments, the method includes: assigning, to each of a plurality of comments, a respective topic vector of length k; determining whether a largest element of the topic vector of a first comment of the plurality of comments exceeds a weight threshold; in response to the determining that the largest element exceeds the weight threshold, classifying the first comment into a first topic, of k topics, the first topic corresponding to the position, in the topic vector, of the largest element of the topic vector; calculating a first sentiment score; calculating an average sentiment score, based in part on the first sentiment score; determining whether the average sentiment score meets a criterion; and in response to the determining that the criterion is met, generating a control signal, the control signal including a message related to the first topic.
1. A method for issuing control signals in response to sentiment, the method comprising:
assigning, to each of a plurality of comments, a respective topic vector of length k, k being a positive integer;
determining whether a largest element of the topic vector of a first comment of the plurality of comments exceeds a weight threshold;
in response to the determining that the largest element exceeds the weight threshold, classifying the first comment into a first topic, of k topics, the first topic corresponding to the position, in the topic vector, of the largest element of the topic vector;
calculating a first sentiment score for the first comment;
calculating an average sentiment score, based in part on the first sentiment score;
determining whether the average sentiment score meets a criterion; and
in response to the determining that the average sentiment score meets the criterion, generating a control signal, the control signal comprising a message related to the first topic.
2. The method of claim 1 , further comprising:
classifying a second comment into the first topic; and
calculating a second sentiment score for the second comment.
3. The method of claim 2 , further comprising displaying a graph of sentiment as a function of time for the first topic, the graph including:
the first sentiment score at a time associated with the first comment, and
the second sentiment score at a time associated with the second comment.
4. The method of claim 1 , wherein the assigning, to each of the plurality of comments, of the respective topic vector of length k, comprises fitting the comments with a topic model.
5. The method of claim 4 , wherein the topic model is Latent Dirichlet Allocation.
6. The method of claim 5 , wherein k is between 30 and 75.
7. The method of claim 5 , wherein the weight threshold is between 0.1 and 0.6.
8. The method of claim 1 , wherein the calculating of the first sentiment score for the first comment comprises calculating the first sentiment score with VADER.
9. A system for analyzing sentiment, the system comprising a processing circuit configured to:
assign, to each of a plurality of comments, a respective topic vector of length k, k being a positive integer;
determine whether a largest element of the topic vector of a first comment of the plurality of comments exceeds a weight threshold;
in response to the determining that the largest element exceeds the weight threshold, classify the first comment into a first topic, of k topics, the first topic corresponding to the position, in the topic vector, of the largest element of the topic vector; and
calculate a first sentiment score for the first comment;
calculate an average sentiment score based in part on the first sentiment score;
determine whether the average sentiment score meets a criterion; and
in response to the determining that the average sentiment score meets the criterion, send a message related to the first topic.
10. The system of claim 9 , wherein the processing circuit is further configured to:
classify a second comment into the first topic; and
calculate a second sentiment score for the second comment.
11. The system of claim 10 , wherein the processing circuit is further configured to display a graph of sentiment as a function of time for the first topic, the graph including:
the first sentiment score at a time associated with the first comment, and
the second sentiment score at a time associated with the second comment.
12. The system of claim 9 , wherein the assigning, to each of the plurality of comments, of the respective topic vector of length k, comprises fitting the comments with a topic model.
13. The system of claim 12 , wherein the topic model is Latent Dirichlet Allocation.
14. The system of claim 13 , wherein k is an input parameter to the topic model.
15. The system of claim 14 , wherein k is between 30 and 75.
16. The system of claim 13 , wherein the weight threshold is between 0.1 and 0.6.
17. The system of claim 9 , wherein the calculating of the first sentiment score for the first comment comprises calculating the first sentiment score with VADER.
18. A non-transitory computer readable medium upon which are encoded instructions that, when executed by a processing circuit, cause the processing circuit to:
assign, to each of a plurality of comments, a respective topic vector of length k, k being a positive integer;
determine whether a largest element of the topic vector of a first comment of the plurality of comments exceeds a weight threshold;
in response to the determining that the largest element exceeds the weight threshold, classify the first comment into a first topic, of k topics, the first topic corresponding to the position, in the topic vector, of the largest element of the topic vector;
calculate an average sentiment score for a plurality of comments including the first comment;
determine whether the average sentiment score meets a criterion; and
in response to the determining that the average sentiment score meets the criterion, send a message related to the first topic.