What it does
In social media platforms such as Twitter, hashtags have evolved into their own unique discourse for categorizing & promoting ideas. Hashtags connect communities and allow organizations to spread ideas with a larger audience and connect with current events. Within this discourse, we have also observed the phenomenon of hashtag hijacking, which presents both an opportunity and a risk.
As an opportunity, popular trends have potential to engage with a large number of people. By assisting organizations in tying in ideas to the relevant conversation, our application can assist in promoting an organization. Tying our solution to GIS allows us to provide them with an entrypoint to local marketing.
As a risk, hashtags can be taken out of context for irrelevant use or abuse. Organizations may adopt a hashtag in order to leverage a wider audience, even if their message is not aligned to the hashtag. Our platform provides textual and geographic analysis can help identify this abuse.
How we built it
In the web interface, we stream live updates of tweets and symbolize according to the polarity value. We also include a list of hashtag currently trending on twitter.
Additionally, we have incorporated analysis of the text of tweets to compare incoming tweets with the same hashtag. We calculate a relevance score based on both the location of the tweet as compared to the current extent of that hashtag's impact area and the content of the tweet based on a dictionary developed on historical data of tweets using that hashtag. Combining the location data and the tf-idf score, we provide insight to whether a tweet is following the general conversation of the hashtag or if it can be detected as an instance of hashtag hijacking.
Challenges we ran into
Our challenges are in connecting the multiple components we have developed into the real time environment of the GeoEvent Server.
We developed the text and location analysis for the relevance score based on select hashtags and based on historical data, and eventually would like to connect this to GeoEvent to evaluate tweets as they come in.
We also developed a model using Model Builder to symbolize a hexagon grid based on aggregating the data points and determining the top hashtag in a local region. In this way, our map would reflect the top local trends.
Our Github page
Accomplishments that we're proud of
We are proud of being able to connect the data from the Twitter Streaming API into our app using GeoEvent Server, getting real time updates to our trend data.
What we learned
What's next for Hashmappers
Our next step would be making final connections to tools we have developed, such as connecting the hexagon symbolization to the GeoEvent processor for real time changes based on trends. We'd also like to connect the relevance scores to GeoEvent so we can build dictionaries on new hashtags in real time and compare incoming tweets to these dictionaries as they arrive.