WOLFRAM

Wolfram Language & System Documentation Center

"FacebookTopic" (Built-in Classifier)

Determine which topic a Facebook post is about.

Classes

Details

  • This classifier attempts to infer the topic of a text.
  • This classifier has been trained on Facebook posts, but can be used for other texts as well.
  • The input text should typically be one or a few sentences.
  • This classifier assumes the topic of the text is unique. The probabilities reflect the belief in these topics, not the proportion of topics.
  • The current version only works for the English language.

Examples

open all close all

Basic Examples  (2)

Use the "FacebookTopic" built-in classifier to detect the topic of a Facebook post:

Wolfram Language code: Classify["FacebookTopic", "I love eating carrots in the morning"]

Classify multiple examples:

Wolfram Language code: Classify["FacebookTopic", {"I bought a new computer", "happy birthday!", "this skirt looks nice"}]

Unrecognized topics or languages will return Indeterminate:

Wolfram Language code: Classify["FacebookTopic", {"what is this topic?", "qwe niowqe mwoei!"}]

Obtain the probabilities for the 3 most likely classes:

Wolfram Language code: Classify["FacebookTopic", "Let's celeberate your new job", "TopProbabilities" -> 3]

Obtain a ClassifierFunction for this classifier:

Wolfram Language code: c = Classify["FacebookTopic"]

Apply the classifier to a list of texts:

Wolfram Language code: c[{"what have you been doing with your life lately", "Let's celeberate your new job"}]

Scope  (1)

Load the ClassifierFunction corresponding to the built-in classifier:

Wolfram Language code: c = Classify["FacebookTopic"]

Obtain the possible classes using Information:

Wolfram Language code: Information[c, "Classes"]

Options  (3)

ClassPriors  (1)

Use a custom ClassPriors to restrict the possible outputs:

Wolfram Language code: Classify["FacebookTopic", "Let's celeberate your new job", ClassPriors -> <|"SpecialOccasions" -> 0.5, "Leisure" -> 0.5|>]

IndeterminateThreshold  (1)

Use a custom IndeterminateThreshold:

Wolfram Language code: Classify["FacebookTopic", "what have you been doing with your life lately", IndeterminateThreshold -> .5]

UtilityFunction  (1)

Obtain the utility function of the classifier:

Wolfram Language code: utility = Information[Classify["FacebookTopic"], "UtilityFunction"];

Modify this utility function to penalize "Weather" being misclassified as "FoodAndDrink":

Wolfram Language code: utility["Weather", "FoodAndDrink"] = -10.;

Classify the post using this new utility:

Wolfram Language code: Classify["FacebookTopic", "What does it mean when deserts freeze?", UtilityFunction -> utility]

Compare to the result with the default utility:

Wolfram Language code: Classify["FacebookTopic", "What does it mean when deserts freeze?"]

Neat Examples  (1)

Obtain different possible topics with their corresponding probabilities:

Wolfram Language code: topics = Classify["FacebookTopic", "what have you been doing with your life lately", "Probabilities"];

Visualize the result using WordCloud:

Wolfram Language code: WordCloud[topics]