Data mining techniques and applications

Computer Science

Data Mining in the Film & Media Industry

Motion pictures, cinema, or films, which are a leading form of entertainment in the visual media industry, is one of the most powerful, influential, and lucrative industries in the world and in modern history. For Americans, it is a common occurrence, for example, to read or hear news reports regarding the huge sums of revenue that films generate in their opening weekends or in the overall run they have in the theaters. The film industry, as part of the greater visual media industry, continues to make strides in technological development and creative innovation, which often translates into large sums of money, or at least impressive sums of money for the amount spent toward the budget and promotion. The film industry is not an industry that has yet successfully used data mining, data analytics, or predictive analytics in a noticeable way. The paper explores the nominal presence of data mining in the film industry right now. The paper also examines and speculates upon other areas within this industry where data mining would prove most useful. Data mining with respect to the film industry is still very new, even when considering how new data mining is generally. It is one of the most prominent industries in the global market that has yet to fully embrace and use data mining to its advantage. The paper will attempt to provide insight into the areas where data mining in film and visual media have emerged as well as offer recommendations toward the applications of data mining in areas that have yet to be considered worthy or valuable of attention and to be data mined.

When considering the relative modern history of the world, (say, the past two or three hundred years), photography and motion pictures have had unbelievable impacts upon human beings. Within the past few decades, advances in information and digital technologies have pushed film and media into new realms, achieving feats that were once only in the imaginations of filmmakers and media producers. There are some media theorists, critics, philosophers, etc. that claim that in the 21st century, a "pure" film no longer exists. Movies we watch now are amalgamations of various forms of visual media pieced and layered together into what audiences believe is a film, but when we pull back the curtain, so to speak, we can see that film & media production is now an attractive layering of graphics, animation, visual effects, special effects, and footage captured by film cameras, video cameras, and even cameras meant for still photography. Directors, producers, and editors, piece and layers all of these elements together into films American and global audiences have become accustomed to, often without noticing the changes to filmmaking consciously.

The earliest uses for data analytics in the film industry were with respect to ratings. Many television viewers and production executives are aware of what is known in America as Nielsen Ratings. Essentially, Nielsen Ratings measure which shows are watched the most and which shows are watched by which specific audience demographics. As one may imagine, this kind of data would prove useful to many people throughout the film industry, yet this data was often restricted to those at the topic of the media foodchain -- studio and network executives, network and film studio heads, and other high level media professionals. In the 1990s and into the 21st century, another prominent type of data collected was again, related to audiences. The film industry began to examine and question audience members with greater scrutiny and detail. This approach to studying audiences was not welcomed at first and not too many people could see the utility to the early attempts to data mine audience preferences, tastes, and choices in film. Some film professionals immediately saw the use for data mining audiences, and still the access to the data was quite restricted and not regarded as overwhelmingly valuable. As the 21st century persisted, data mining, data science, data analytics, and big data became more commonly used terms and practices. As other professionals make use of data mining toward the more in-depth exploration and innovation of various industries, there is greater credibility toward the kinds of insights and applications for the film industry. Still, this industry treads lightly in this area waiting for brave pioneers to step forward to demonstrate what data mining can do and how it can help the film industry in ways that many professionals in this field have yet to consider.

And in a Wired feature that went up this week, reporter Roberto Baldwin takes a closer look at whether Netflix's use of math and information on past streaming habits is really revolutionizing the way TV series are developed. While networks traditionally order a show based on whether it likes a pilot, Netflix ordered two full seasons (26 episodes) of House of Cards without seeing a single scene. It reportedly bid more than $100 million to secure first rights to the show, outbidding HBO and AMC because it is utterly convinced the show will be a big hit. Why? Because it is counting on data mining and algorithms to provide an edge. The company knows how many people are watching Kevin Spacey and David Fincher movies and it knows how many viewers watch political thrillers. If that audience is large enough, getting exclusive access to House of Cards makes sense. (Willmore, 2012)

In this case, Netflix has altered its decision-making process and choice of investments based on data mining. This is quite new. Netflix is a service that provides users access to DVDs and streaming video of a massive collection of television programming and films around the world and from the earliest moments of film history to projects that are brand new. Netflix is a service provider that is behaving like a premium television network. This quotation states how Netflix rivaled HBO and AMC, channels that offer some of the most provocative, creative, and entertaining programming in the United States, much of which ends of getting distributed around the world. The use of data mining in the film & media industry has effected big changes in Netflix. The use of data mining seems to have altered the course or trajectory of this company's destiny. It is able to enter different markets and rival some of the best networks with confidence. Data mining has empowered Netflix to take leaps and bounds in a way that other television networks and film studios are hesitant.

Netflix uses a piece of software composed of several algorithms called CineMatch. It is how Netflix data scientists mine the data of the company, primarily how people rate films they have seen so that Netflix can more accurately predict other films and shows that users will enjoy based on their ratings and tastes. The approaches Netflix uses to mine film and user data are association rule learning, classification, and regression. Perhaps these same approaches can be applied to data mine the film industry overall.

This is an early step that has profound potential to be the beginning of a trend that could radically change the film & media landscape in the United States, but also in other countries with booming film industries such as China, and India. India, for example, has long since held the record for producing the greatest number of films per country out of Bollywood. The Chinese cinema additionally has a long history of progressive methods of production, special effects, camera & editing techniques. This is another country that produces a great amount of films per year, and continues to do so for years. If there are just these small changes happening in the United States, we can only imagine what kind of changes will happen in other countries who may have a more open and progressive attitude toward data mining with respect to film & media. As data mining spreads the film industries around the world, there will be more resources and visualizations to assist media professionals in understanding trends and patterns in spending, content, distribution, marketing, and even areas such as casting and elements of design. Producers, executives, and the creative members of the crew

The film industry, in its baby steps toward embracing and utilizing data mining, analytics, and science on itself, has looked to other forms of media to help it transition into a more data dependent industry. For example, seeing positive evidence from the gaming industry, mobile applications, and social media, the film industry shows signs of slowing showing confidence in data mining.

The vast amount of information that exists in social media about films, actors, and third-party product tie-ins can provide an instant barometer to how people are reacting. Being able to collect and crunch that data quickly means a company can craft a response and alter a marketing campaign in a way that results in business success. "The real time aspect of social media is especially critical to the movie industry," said Abe Kazemzadeh, CTO of the Annenberg Lab. Social sentiment analysis can be decrypted to show geographical, demographic, and use case data that can be correlated with box office figures, he added. (Karl, 2012)

Social media is not the same as film, nor it is directly a part of the film industry, yet film is an extremely popular subject of discussion on social media platforms. By data mining social media content that is related to film, film industry professionals have the opportunity to gain valuable insights into what is working and what is not. In a way, data mining social media content regarding films is another type of audience analysis or Nielsen ratings that the film industry does not have to seek out or gather itself; social media users are talking about film already without any prodding or motivation from the film industry specifically. Social media has great potential to change how films are marketed, distributed, and produced. Data mining social media on behalf of the film industry is a practice that more industry professionals should make a habit because the data found is invaluable and because at this point, data mining is still cutting edge. Film professionals who practice data mining are ahead of the curve; those who stay ahead of the curve have the power to influence the industry for years to come.

On an individual level, data mining has productive implications for the struggling or newcoming filmmaker. The film world is a small world; it is hard for new people to break in. When people pitch their scripts and ideas to producers or executives, it is a daunting process that comes with a lot of harsh rejection. Besides having a strong idea that is executed well, individual filmmakers would do well to invest in data mining for their own ideas. After all, how many filmmakers do we think there are that go into a pitch meeting with data visualizations that illustrate or distinctly prove how the genre of their screenplay hits or attracts key demographic or niche demographics? How many filmmakers know how to use data mining to raise funds to finance their productions? How many filmmakers provide data visualizations that show the breakdown of how much of the production budget is dedicated to particular aspects of production or how much time/labor is dedicated to certain areas of the production? The answer is very few now, but likely to be very many in the very near future. That kind of information could additionally assist executives making better hiring decisions and better inform their investment decisions. A person who enters a producer's office with a great idea, presents it well, and that also included data to back up their claims that a specific type of film appeals to specific audiences and can be done within a specific budget, would have a greater likelihood to receive financing and a greenlight for their project than someone with the same ideas and talents who did not also have data to supplement their pitch.