SIM/script:
Invest now in a tool that will help content providers meet the challenges ahead.
Be among our first clients and have your content evaluated against best-of-class assets of its format.
Some background
Michael Esser Development is a sole proprietor enterprise, an established consulting
company that specializes in advising European customers on entering the US market.
In Germany, Switzerland and Italy, MED is known as a trustworthy partner when it
comes to preparing projects for a presentation in the US, preparing or producing the
necessary materials, and preparing and coaching staff for pitching at studios,
networks, agencies and streamers.
In Los Angeles, MED has established itself as a reliable connection to European media
companies. MED is known as a reputable provider of interesting materials and
development projects and as a matchmaker between US companies and companies,
especially in Germany, who want to cooperate in the development of scripted projects
for the international TV market .
MED intends to develop SIM_script, an NLP-based software service for content
providers and producers. The development of SIM_script takes place in a company
that is founded for this purpose in the form of an LLC.
The Project Origin
Michael Esser, writer, development consultant and media scientist, has received his
Ph.D. in Media sciences at TU Berlin, Germany. He researched methods to effectively
support screenwriters to achieve output of a consistent quality according to a set of
specifications, such as genre, tone, format, and style. In a three-year on/offline field
experiment he compared a set of strategies to deliver the best possible contribution to
existing popular TV formats with minimal effort.
Sergei Monakhov, digital linguist at TU Jena, has developed an NLP algorithm that
ranks social media posts according to the probability of this posts to be authored by
trolls. The algorithm looks at the text only, not at metadata like address or pathway.
The algorithm Sergei developed shows a high detection rate even on small numbers of
samples.
In April 2020 Michael heard an interview with Sergei on BBC’s program Tech Tent.
Sergei explained the functionality of his algorithm, and how it could detect social media
postings by internet trolls through linguistic patterns. Something clicked with Michael.
Could the algorithm also detect patterns in screenplays of a given genre, or episodes
of a specific TV show?
Together, Sergei and Michael have realigned the troll detection tool. They have trained
the algorithm to recognize patterns in a set of scripts. These sets can be the most
successful scripts of a specific genre or scripts that belong to a specific TV show.
The result: SIM_script
– SIM script develops parameters typical for a given genre or show format.
– SIM script evaluates new content on the basis of those linguistic patterns.
– SIM script ranks new content according to the evaluation and thus creates
meaningful results.
NNM – Technical Background
In a study in 2020 Sergei Monakhov proposed to understand Troll Writing as a Linguistic Phenomenon. Monakhov achieved an accuracy score of 91% in classifying tweets as originating from trolls by means of fairly simple neural network modeling (NNM), thus showing that identifying troll writing is not difficult provided there is a large corpus of examples. In the tweets that—for the lack of a better term—will be referred to as ‘genuine’, with every new post from the author, the probability of meeting a word that has already been used should increase. In contrast, in troll writing, with every new tweet, the probability of meeting a word that has already been used should continuously decrease or stay constant, at least until a certain cut-off point is reached. This is inevitable, provided that the initial hypothesis of trolls being forced to use target words in a variety of different contexts is justified.
We made the simple assumption that also the quality of a screenplay, seen as authenticity of intent, should be detectable in the same way.
For reference:
Understanding Troll Writing as Linguistic Phenomenon – Sergei Monakhov
[download: Understanding_Troll_Writing_as_a_Linguis.pdf]
Early Detection of Internet Trolls – Sergei Monakhov
[download: Early_detection_of_internet_trolls_Intro.pdf]
Status quo and further steps
SIM/script has been in development since May 2020. The first investment was made by Michael Esser and Sergei Monakhov.
The underlying algorithm had passed through two intensive rounds of testing based on 60 scripts of two TV series formats and 80 screenplays of movies of the genre Horror.
The Product Pitch (available separately from this Business Plan) gives a detailed overview about the development process, describes the tests in detail, and provides test results.
[download: SIM script pitch – Product Pitch v2.00-02-25-2021.pdf]