DETERMINING USER SENTIMENT IN CHAT DATA
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for receiving a message authored by a user, determining, using a first classifier, that the message contains at least a first word describing positive or negative sentiment and, based thereon, extracting, using a first feature extractor, one or more features of the message, wherein each feature comprises a respective word or phrase in the message and a respective weight signifying a degree of positive or negative sentiment, and determining, using a second classifier that uses the extracted features as input, a score describing a degree of positive or negative sentiment of the message, wherein the first feature extractor was trained with a set of training messages that each was labeled as having positive or negative sentiment.
1 . A computer-implemented method comprising:
extracting, using a first feature extractor, one or more first features from a message, wherein each first feature comprises a respective word or phrase in the message and an associated weight signifying a degree of positive or negative sentiment;
extracting, using a second feature extractor, one or more second features from the message, wherein each second feature comprises a distance between a first word and a second word in the message, wherein the first word comprises at least one of a conditional word and an intensifier word, and wherein the second word comprises at least one of a positive sentiment and a negative sentiment; and
determining a score describing a degree of positive or negative sentiment of the message based on output of a trained classifier, wherein the extracted first and second features are provided as input to the classifier.
2 . The method of claim 1 , wherein the classifier was trained with features extracted by the first and second feature extractors from a set of training messages.
3 . The method of claim 1 , wherein the first feature comprises an emoticon, an emoji, a word having a particular character in a correct spelling form of the word that is repeated consecutively one or more times, a phrase, an abbreviated or shortened word, or a text string with two or more consecutive symbols.
4 . The method of claim 1 , wherein extracting, using the first feature extractor, one or more first features from the message comprises using an artificial neural network feature extractor to extract the features.
5 . The method of claim 1 , wherein the classifier comprises a naive Bayes classifier, a random forest classifier, or a support vector machine classifier.
6 . The method of claim 1 , further comprising:
extracting, using a third feature extractor, one or more third features of the message, wherein each of the extracted third features comprises:
(i) two or more consecutive words that describe positive or negative sentiment;
(ii) a count of words, symbols, biased words, emojis, or emoticons; or
(iii) a word having a particular character in the word's correct spelling form that is repeated consecutively one or more times.
7 . A system comprising:
one or more computers programmed to perform operations comprising:
extracting, using a first feature extractor, one or more first features from a message, wherein each first feature comprises a respective word or phrase in the message and an associated weight signifying a degree of positive or negative sentiment;
extracting, using a second feature extractor, one or more second features from the message, wherein each second feature comprises a distance between a first word and a second word in the message, wherein the first word comprises at least one of a conditional word and an intensifier word, and wherein the second word comprises at least one of a positive sentiment and a negative sentiment; and
determining a score describing a degree of positive or negative sentiment of the message based on output of a trained classifier, wherein the extracted first and second features are provided as input to the classifier.
8 . The system of claim 7 , wherein the classifier was trained with features extracted by the first and second feature extractors from a set of training messages.
9 . The system of claim 7 , wherein the first feature comprises an emoticon, an emoji, a word having a particular character in a correct spelling form of the word that is repeated consecutively one or more times, a phrase, an abbreviated or shorted word, or a text string with two or more consecutive symbols.
10 . The system of claim 7 , wherein extracting, using the first feature extractor, one or more first features from the message comprises using an artificial neural network feature extractor to extract the features.
11 . The system of claim 7 , wherein the classifier comprises a naive Bayes classifier, a random forest classifier, or a support vector machines classifier.
12 . The system of claim 7 , wherein the operations further comprising:
extracting, using a third feature extractor, one or more third features of the message, wherein each of the extracted third features comprises:
(i) two or more consecutive words that describe positive or negative sentiment;
(ii) a count of words, symbols, biased words, emojis, or emoticons; or
(iii) a word having a particular character in the word's correct spelling form that is repeated consecutively one or more times.
13 . An article comprising:
a non-transitory computer storage medium having instructions stored thereon that when executed by one or more computers cause the computers to perform operations comprising:
extracting, using a first feature extractor, one or more first features from a message, wherein each first feature comprises a respective word or phrase in the message and an associated weight signifying a degree of positive or negative sentiment;
extracting, using a second feature extractor, one or more second features from the message, wherein each second feature comprises a distance between a first word and a second word in the message, wherein the first word comprises at least one of a conditional word and an intensifier word, and wherein the second word comprises at least one of a positive sentiment and a negative sentiment; and
determining a score describing a degree of positive or negative sentiment of the message based on output of a trained classifier, wherein the extracted first and second features are provided as input to the classifier.
14 . The article of claim 13 , wherein the classifier was trained with features extracted by the first and second feature extractors from a set of training messages.
15 . The article of claim 13 , wherein the first feature comprises an emoticon, an emoji, a word having a particular character in a correct spelling form of the word that is repeated consecutively one or more times, a phrase, an abbreviated or shorted word, or a text string with two or more consecutive symbols.
16 . The article of claim 13 , wherein extracting, using the first feature extractor, one or more first features from the message comprises using an artificial neural network feature extractor to extract the features.
17 . The article of claim 13 , wherein the classifier comprises a naive Bayes classifier, a random forest classifier, or a support vector machines classifier.
18 . The article of claim 13 , wherein the operations further comprise:
extracting, using a third feature extractor, one or more third features of the message, wherein each of the extracted third features comprises:
(i) two or more consecutive words that describe positive or negative sentiment;
(ii) a count of words, symbols, biased words, emojis, or emoticons; or
(iii) a word having a particular character in the word's correct spelling form that is repeated consecutively one or more times.
19 . The method of claim 1 , wherein the first word comprises the intensifier word.
20 . The system of claim 7 , wherein the first word comprises the intensifier word.