Memes are becoming premier sources of information, especially for Generation Z. People find enjoyment both when looking at funny memes themselves and sharing these memes with their friends (and thus hopefully making their friends laugh). These memes are transmitted through channels such as Facebook Messenger, email, social media, text, meme apps, etc.
Since a meme is essentially just a picture with some text on it, it is difficult to classify. Thus, it often takes a person much “meme hunting” before finding one that satiates his/her sense of humor, and these “senses of humor” are extremely difficult to classify in simple terms. This presents a problem of inefficient searching for enjoyable memes and content. The main services on which users find memes are Facebook, Twitter, Instagram, and Snapchat, all of which use feed systems that are filled with content posted and shared by other accounts. People spend millions of hours per day swiping, scrolling, liking, and commenting on memes every day. However, people generally have short attention spans when browsing content and are prone to get bored, distracted, and / or frustrated while searching for memes. With advanced technology and Internet browsing available at their fingertips, people may just give up searching for memes and engage in another activity if they are unable to find ones that appeal to their unique senses of humor.
Our idea is to use machine learning and clustering around “senses of humor” to curate memes for individuals, creating a “Tinder for Memes”. This solution offers both time-saving and more concentrated enjoyment. This would cut down time on “meme hunting” by going through individuals’ preference data (obtained through a swiping mechanism) to present only the memes that the user would find most entertaining. Memes are thus better tailored to each user over time as profile feeds become increasingly personalized through increasing user input in the form of “swipes”. This will thus cut down on searching time and provide more enjoyable memes, on average, that can then be shared with friends. The user interface will consist of a personalized feed of memes and options to either “like” (swipe right) or “dislike” (swipe left) on each image, and a suggestion algorithm will utilize clustering data. This will initially be developed to be a web application but hopefully a smartphone application eventually.
There are no services or applications on the market that offer self-improving meme distribution, as most people have resorted to more-tailored Facebook groups and forums. That being said, our value-add is partially gained from individuals who may like many types of memes, and an algorithm could be designed to hit these many “senses of humor” well.
There are multiple avenues for the business model. First, memes offer huge potential content opportunities for big brands, and not many marketing departments utilize memes. The machine learning and swiping mechanism process could be used to segment customers for advertisers and thus provide more targeted advertising for brands. If the application was to be run as a standalone business, it could generate revenue through in-app advertisements (possibly meme advertisements) and through paid subscriptions that eliminate advertisements and possibly add functionalities. For example, after every 10 memes that are shown, an advertisement could appear on the screen that could be “left-swiped” to toss away or “right-swiped” to be viewed.
Possible additional implementations include implementing Facebook and Twitter API (if possible) to gather initial user data to complement the curating process. Linking the application to Facebook will allow for easy transfer of useful data to the application and user profiles. Eventually, we hope to allow users to post their own memes and track likes / dislikes, send memes to and message friends, and search for memes. Users will have a database in which they are able to access all memes they previously “liked”.