Identifying Purchase Intent in Social Posts

1 . A computer-implemented method comprising:

receiving a topic;

collecting social posts to one or more social networks that are related to the topic; and

identifying, from the collected social posts, one or more purchase intent posts, the purchase intent posts expressing purchase intent towards the topic.

2 . The computer-implemented method of claim 1 , wherein the identifying the one or more purchase intent posts comprises computing a confidence level for each of the collected posts, and identifying the one or more purchase intent posts based on the confidence level.

3 . The computer-implemented method of claim 2 , further comprising sorting the collected posts based on the confidence level of each post.

4 . The computer-implemented of claim 2 , where the identifying further includes identifying the one or more purchase intent posts based on the confidence level of the one or more purchase intent posts exceeding a confidence threshold.

5 . The computer-implemented method of claim 2 , wherein the confidence level is computed by analyzing the collected posts to determine whether each post includes features indicative of purchase intent.

6 . The computer-implemented method of claim 5 , wherein the features indicative of purchase intent include one or more of an object of desire related to the topic, an action verb depicting purchase intent towards the object of desire, or a self-reference to a user.

7 . The computer-implemented method of claim 5 , wherein the features indicative of purchase intent further include one or more of a named entity or a positive sentiment word.

8 . The computer-implemented method of claim 1 , further comprising:

separating the one or more purchase intent posts from one or more non-purchase intent posts; and

causing display of representations of the one or more purchase intent posts and the one or more non-purchase intent posts in a user interface.

9 . The computer-implemented method of claim 8 , wherein the separating and causing display is performed in real-time as the social posts are being posted to the one or more social networks.

10 . A computer-implemented method for building a purchase intent model comprising:

analyzing a training corpus of annotated social posts by linguistically parsing text of the annotated social posts to locate features indicative of purchase intent; and

building the purchase intent model by passing the features indicative of purchase intent to a machine-learning model.

11 . The computer-implemented method of claim 10 , wherein the features indicative of purchase intent include one or more of an object of desire, an action verb depicting purchase intent towards the object of desire, a self-reference to a user, a named entity, a positive sentiment word, or bag-of-words based delta-term frequency-inverse document frequency (TF-IDF) vectors.

12 . The computer-implemented method of claim 10 , wherein the features indicative of purchase intent include purchase intent words (PI words) and non-purchase intent words (non-PI words).

13 . The computer-implemented method of claim 12 , wherein the PI words and the non-PI words are determined by:

parsing the text of the annotated social posts to locate verbs and nouns; and

scoring the located verbs and nouns based on a comparison of a number of times that the located verbs and nouns occur in annotated social posts expressing purchase intent and a number of times that the located verbs and nouns occur in annotated social posts not expressing purchase intent.

14 . The computer-implemented method of claim 12 , further comprising expanding a number of PI words and non-PI words recognized by the purchase intent model using one or more external databases.

15 . The computer-implemented method of claim 14 , wherein the one or more external databases comprise at least one of Wordnet or Freebase.

16 . One or more computer-readable storage media comprising instructions stored thereon that, responsive to execution by a computing device, cause the computing device to perform operations comprising:

receiving a topic;

collecting social posts to one or more social networks that are related to the topic; and

identifying, from the collected social posts, one or more purchase intent posts, the purchase intent posts expressing purchase intent towards the topic.

17 . The one or more computer-readable storage media of claim 16 , wherein the identifying the one or more purchase intent posts comprises:

computing a confidence level for each of the collected posts; and

identifying the one or more purchase intent posts based on the confidence level of the one or more purchase intent posts exceeding a confidence threshold.

18 . The one or more computer-readable storage media of claim 17 , wherein the confidence level is computed by analyzing the collected posts to determine whether each post includes features indicative of purchase intent.

19 . The one or more computer-readable storage media of claim 18 , wherein the features indicative of purchase intent include one or more of an object of desire related to the topic, an action verb depicting purchase intent towards the object of desire, a self-reference to a user, a named entity, or a positive sentiment word.

20 . The one or more computer-readable storage media of claim 16 , wherein the instructions, responsive to execution by the computing device, cause the computing device to perform operations further comprising:

separating the one or more purchase intent posts and one or more non-purchase intent posts; and

causing display of representations of the one or more purchase intent posts and the one or more non-purchase intent posts in a user interface.