Analysing emojis can uncover meaning and sentiment in ways regular text analytics cannot. So this was the main idea to introduce sentiment data into the Emoji-Heatmapper app. LokLak Search API has features such as classification and categorization of tweets. The emotions, for instance, can be joy, anticipation, sad etc.

So, in the Emoji-heatmapper app, I am displaying the occurrence of emojis on the map according to the location traced and also the sentiment related to the emoji i.e., search query as follows:

How to get the sentiment data:

One should simply enter the emoji into the search box for the results. The following code shows part of the LokLak Search API results (JSONObject):

      "hashtags_count": 0,
      "classifier_emotion": "anger",
      "classifier_emotion_probability": 
      1.2842921170985733E-9,
      "classifier_language": "english",
      "classifier_language_probability": 
      1.4594549568869297E-8,
      "without_l_len": 49,
      "without_lu_len": 49,
      "without_luh_len": 49,

I am using the above field name ”classifier_emotion” to display the results.

Till here getting the data relevant to query part is done. Next, the classifier_emotion of each tweet containing the query is collected into an array and sorted to get a unique list.

var emotion = [];
var emotions_array = [];

for (var i = 0; i  <  tweets.statuses.length; i++) {
    if (tweets.statuses[i].classifier_emotion) {
        emotion = tweets.statuses[i].classifier_emotion;
        emotions_array.push(emotion);
    }

    emotions_array.sort();
    emotion_array = jQuery.unique( emotions_array );

Loading the Sentiment data onto the Screen

When the query has a single emotion, or if multiple emotions or no emotions. These use cases/situations are displayed as follows:

Fig: Single Emotion

Fig: Multiple Emotions

Fig: No Emotions data

The code which creates the data dynamically on the output screen is as follows:

//Loading the sentiment
$(document).ready(function() {
    var listItems= "";
    if (emotions_array.length == 0) {
        listItems = "No Sentiment data is available for " + query
    }
    if (emotion_array.length == 1) {
        listItems += "<h3> Sentiment of " + query + " is ";
    } else if (emotion_array.length > 1) {
        listItems += "<h3> Sentiments of " + query + " are ";
    }

    var emotion_data = emotion_array.join(", ") + "."
    listItems += emotion_data + "</h3>"

    $("#sentiment").html(listItems);
});

Conclusion

The Emoji-Heatmapper app displays the sentiment data of the query being searched for which populates data dynamically using LokLak Search API.

Resources

• Emoji-Heatmapper App, try it out here: http://apps.loklak.org/emojiHeatmapper/
• Source Code: https://github.com/fossasia/apps.loklak.org/tree/master/emojiHeatmapper
• Search API: http://api.loklak.org/