Pressum

Using AI to connect Indians with stories they care about. 

Historically, Indians have held the collective power to bring systemic change to their country which is seen from high profile cases like Nirbhaya that saw amendments to Article 375 of the Indian constitution since 1983. But they have very rarely capitalized on this with only 17 major movements in the past 7 years. It is clear that Indians want social change as seen by Parul Bansan’s 2013 paper of Indian youth and social identity & are undoubtedly willing to fight for it. 

However, there are rarely clear rallying points that can be used to springboard systemic change at various levels. This is a grave flaw in any republic, for democracy dies in darkness. Not having clear rallying points not only prevents widespread systemic change that should be commonplace in NICs such as India, but also impacts victims of the status quo. In numbers, this is 28,105 yearly rape victims, this is the 71% of elders who are abused in India, this is 15 cr. children who are out of school and more. Such issues prevent us from building inclusive societies where all feel safe and encourage unaccountable systems that fail to provide justice & security. In doing so, they prevent us from achieving goal 16 of the UN SDGs.

It’s not a lack of cases that creates such a problem as we have encountered many as severe as Nirbhaya during our research through the archives of popular journalism agencies such as ‘The Hindu’. Rather, these stories are buried in archives that though accessible, are unnoticed until specific attention is paid. This is the root cause of this flaw, & it’s main contributing factor. Hence, we identify the lack of a platform to uncover these unheard stories as a problem.

Pressum (Latin for concise) aims to be the go-to website that provides Indian communities with unheard news that matters. 

We seek to do this by dividing website members into separate stakeholder groups that can be given relevant news they can act on (along with other news on less relevant topics, but at a lower priority). This news will be drawn from online archives after being sorted by AI models into separate subjects of concern such as senior abuse, rape, etc.

The chosen articles will then be reworded in a concise fashion for ease of understanding and an AI model will work to find any follow-ups to ensure stakeholders are always up to date with cases that concern themselves. This will allow users to understand what is being done about their interests & take action if they feel underserved. Such responses could be organized on the website itself, or with partnerships with Indian charities that focus on protest action.

Our solution divides stakeholders into two distinct categories, victims and respondents. Pressum aims to make unseen news about victims easily accessible to respondents, so some may stand up for the victims and take appropriate action to bring them justice. This could include donations, petition & later on, peaceful protest.  

Hence though this solution is “for” victims, it directly serves respondents - a class of stakeholders that are not yet affected but know that they might be. This includes middle-class retirees who wish to stand up for veteran pensions, young women standing up for corporate acceptance or any other group which has a vested interest in justice. 

These groups are critically underserved due to India’s news landscape, which focuses on TRP ratings & creating dramatic situations before reporting. This has almost disenfranchised them as they see news as a waste of energy due to its unproductive nature, which leads to them taking no action in their interest. Pressum aims to break this cycle and save victims by informing respondents.

As a team, we possess helpful skill sets and experiences that put us in a good position to solve the described problem.

Our team have a diverse set of technical skills as some are fluent in languages such as Java, Python, and C# while others showcase skills through web design in applications such as HTML and CSS. Furthermore, a member of our team has developed machine learning models under the guidance of Nanyang Technological University and Hewlett-Packard Enterprise. This allows us to power the core technologies of our solution.

Each group member also has applicable research experience ranging from media studies, economics, history, and chemistry allowing us to obtain an interdisciplinary understanding of research papers & their components. All group members have contributed to school departmental research at leadership levels and understand the need for teamwork and cohesion in a field as fluid as research. One group member has formerly worked with Learn with Leaders and has won a research grant to further his study of dynamic educational systems.

During an economics research project to measure correlation between news exposure & donations our team explored E-archives from Indian news sites. Throughout the process were exposed to the dark realities of senior and sexual abuse within India in great detail. As observers, we never dived deep into unpopular issues and as such had a very numerical understanding of the problem. However, exploring these archives gave us more knowledge on such issues than we were exposed to in our whole life prior. This was an indication to us that articles are not easily accessible to the public which got us to work on this problem.

Furthermore, we recognize that empathy is the most important qualification a team can possess. It allows us to truly represent our clients & see the logic in their statements when others see fantasies.

 Members of our team have actually been able to relate these gruesome news readings with the unfortunate experiences of their loved ones. In one case, a member’s grandmother was robbed of her family jewels when she was helping a sick sister. As the door was broken into, furniture pushed rigorously, and her legacy of gold for her children was gone in a matter of minutes, there was only going through our heads. What would have happened if she had been there that night?

As problems of senior abuse, rape cases, and other problems, become more prominent, this case of his grandma was just the tip of the iceberg. Ending off, we are a representation of the people we serve as we are among the inhabitants of India and have personally experienced the problem just like everyone, allowing us to look from the perspective of the target audience to solve the problem better.

Understanding strength in diversity, we decided to conduct a highly varied primary research consisting of musicians, artists, doctors, teachers, students, programmers & other professions. We collected twenty-five conversational style interviews each 5-10 minutes long.

 These interviews provided us with valuable insights and displayed expected dissatisfaction on the news industry in India. Most interviews ended with endorsements of our idea; however, descriptions of unusual problems and quick fixes were the most desirable parts of the conversations since they highlighted what normal citizens go through every day to get their news the way they like it. Our most valuable insights included:

• To present issues on our website in the form of stories that are updated rather than articles, to allow people to better connect and understand issues – Most respondents (23) agreed with this principle since they were tried of news that went nowhere.  

• To ensure that the information being published in the website are facts as we are working on sensitive matters – This point was brought up independently by several history teachers (2).

• Teachers in particular (8 out of 9) mentioned how they were avid news readers and actually shared similar experiences with us. Many of them often pondered the implications of such concerning topics and how they often had to research on news sites, rather than it being available to them on their feeds at first glance.

• While some prefer unfocused news that covers a vast of range of topics (3), there are many (22) who are looking for sources that could provide them with better knowledge on these important matters, indicating the demand for such a solution. 

When studying a foreign language, flashcards are highly utilized as they grant sustained exposure in focused areas. This allows learners to improve retention and focus on specific areas that they have identified for any amount of time. Such an approach is effective not only for those working towards fluency but also for students trying to score high, tourists trying to understand basic instructions, academics studying primary sources and so on. Oftentimes, this is better than using textbooks which are rigid, unscalable & outdated even if they provide more information than flashcards at a base level.

This is the philosophy behind Pressum. While news sites represent textbooks, we represent dynamic flashcards. Typically, news is presented in an endless flow and this approach drowns the reader with information. It reduces the attention paid to each story and incentivizes quantity over quality.

We aim to bring change by focusing on impactful stories which are frequently updated to ensure readers know what happens with cases they care about. Traditional news does not focus on this & readers are left to piece together stories themselves. By curating stories for our audiences, we can ensure they’re informed about not only about its base contents but also its history, scope, impact & the action they can take. Such individual focus is what we believe gives our solution the potential to be catalytic. The current news market in India has nothing that is as focused as we aim to make Pressum, it is scattered and overextended & simply cannot narrow its focus as much as we can. This means a new medium such as ours can fill in the vacuum that has undoubtedly risen from decades of such news.

It can hence, generate positive societal impacts by informing, educating and preparing stakeholders who have historically felt disconnected from the aforementioned style of news. 

As a purely web-based solution, Pressum’s goals are different from more physical products. Despite this, we have identified three concise goals that will help us generate catalytic social impact.

-In 4 months, we aim to develop, test and finalize AI models that can identify news as per our previous specifications. During this time, creating a dynamic website will also be prioritized with expected completion.

-Within 6 months, we aim to publish our website and place growth focus on various urban markets in India such as Chennai, Mumbai, Bengaluru & New Delhi. Though 5,000 users throughout India is our goal at this stage, we would prefer if it was concentrated in one or two of the cities mentioned.

Partnerships with local government organizations like police stations, municipal bodies and public libraries along with small local NGOs will be our main growth strategy.

-Within 9 months, we aim to expand our website to 5-10 less significant tier-II urban markets such as Patna, Mangalore, Madurai and Salem & most tier-I cities we haven’t reached yet. 

-At our 12-month anniversary we aim to create a nationwide donations platform for partnered NGOs to provide stakeholders with ways of advancing their interests. We also wish to help them organize peaceful, legal and orderly protests by connecting them to each other & the right authorities (local police). Such peaceful and legal protests are the backbone of Indian democracy and have been upheld by the supreme court of India in 1962, 1978 & 1986. We have set a user goal of 17,000 for this checkpoint. 

By this point, we expect our user count in any one urban market to be sufficient to sustain this. (Though Bengaluru will likely be the first market to experience this, we know it is important to be open to all opportunities, hence growing markets will be given priority even if they are classified as tier-II.)   

Pressum aims to be the modern flashcard wall for news and the technology that empowers it closely resembles this. 

Much like paper is used for flashcards, a dynamic website will be used for Pressum. Though not a software application, it will allow users to access stories in small concise bites like we envision. This website will contain all we need and be the basis of our solution. 

But arranging, writing and maintaining these flashcards is more important than the paper used, and this represents our core technology of Machine Learning utilizing Natural Language Processing (NLP) 

We will use large datasets (100,000 samples plus) to train ML models in categorizing articles into distinct classes, some of which we are interested in. Supervised machine learning will then be used to analyze user activity and identify which users have the most interest in a potential story before notifying them & attaching it to their feed. With arranging and writing covered, maintenance is all that remains. This will be handled by our third ML model which will work to identify any follow up articles which can be curated & linked to the original story to allow ease of access. 

Such a process leads to a platform that users know has stories and information they care about & its near complete automation allows us to focus on improving user experiences and vastly improves scalability.

As a concept, Pressum does not have any current customers. Despite this, we have set clear yearly user goals divided by markets.

Tier I - We aim to serve all 6 of India's tier I cities (Mumbai, Kolkata, Chennai, New Delhi, Bengaluru, Hyderabad) with 2,000 users per city on average. 

Tier II - We aim to serve 10 tier II cities with 500 users per city on average. 

Bengaluru - Though we cannot predict how each market will react to our product, we feel the Bengaluru will be the easiest to crack into, we estimate 5,000 - 7,000 users within one year of our launch. 

Pressum as a concept faces several barriers that may hamper its abilities to bring forth catalytic change. 

• Financial: The technologies highlighted above require significant one-time expenditure that our team cannot justify. This is a result of hardware requirements since new and faster computers will be essential to extract data for use in training & news. After initial investment, monthly fees for software tools and web hosting will further drive up the price, and these will only increase as time goes on.  

• Technical: Though our team has experience in creating datasets and ML models for research, we have never required to create such large datasets. Though using manually correcting data generated spreadsheets and reusing it as training material is possible, it is time consuming & slows our rollout. 

• Legal: Pressum’s work may inadvertently violate the Indian Copyright Act of 1957 by miscrediting articles unintentionally. This requires us to pay dedicated attention to each entry to ensure they are properly credited & do not break IPR. This again, slows Pressum's initial rollout. 

So far, we do not partner with any organization. 

Pressum will make use of two distinct business models for our growth and late/independence stage. Each of these models work to maximize our strengths at both stages. 

• Growth (First year) - Beginning with a team of 4 and possible grant money & partnerships, this stage will see rapid expansion as a priority focus. We will place emphasis on collaboration with local NGOs and conduct more extensive primary research with key stakeholders such as senior citizens & feminists. 

Early on, hardware expenses and one-time software setups will cost around 120,000 INR (1,470 USD) which is 82.1% of our expenditure. Recurring expenditure for software, hosting, domains and so on will come at 21,500 INR (263 USD) or 17.9%. This low starting cost means our product can function with grant money alone for 13 months, however marketing fees are unaccounted for. This means 6-7 months is a more reasonable 

In this time, our main beneficiaries (and customers) will include senior citizens & busy individuals who now have access to relevant news they otherwise wouldn't have had. Our main channels of expansion will include NGO partnership & word-of-mouth promotion. We do not expect any revenue during this period, but surplus gained from grants or donations will be returned if requested or stored for anticipation of greater expenses. Hence, we will adapt a donation model at this stage and focus on expansion & preparation for the next stage. 

• Late - Human capital will retain its role as our most essential resource, as such we might consider expanding this by recruiting like-minded volunteers for 2-3 month internship cycles. We also see expansion of physical capital likely, since better hardware will be required to train AI models at a faster pace. Though this will increase one-time investments, recurring expenses will likely take a greater share of our cost structure as an increase in users will increase hosting fees. 

To match this, we aim to adapt a subscription business model with a spinoff product - Pressum Casual. This new concept will have all the underlying philosophies of the original, however, it will focus on arts, the humanities, history and more lighthearted subjects. Though we will use knowledge from Pressum on this, it will be a completely separate freemium-to-premium (monthly cost to removing ads) tool that will target students & humanities professionals. Surplus revenue generated through this idea will be invested in it, as a strong financial foundation can help Pressum proper remain financially independent as it attracts growing quantities of users. 

Many of our interviewees (20) lacked trust in Indian media due to financial incentives & screen ratings, hence, if Pressum wants to be trusted by its users, it must remain financially independent and unaligned.

Therefore, a combination of donations and grants along with the later sale of a freemium subscription in an offshoot will be used to sustain Pressum. 

• Donations & Grants - Though we aim to reduce our reliance on donations & grants, we will not entirely phase them out. We hope to provide users a platform to donate and aim for a 0.5% donation rate with at least 200 Rs. (2.45 USD) among a potential 17,000 users at the end of year 1. 

17,000 users x 200 Rs. x 0.5% of users = 17,000 INR (around 208 USD) 

Combined with semi-regular grants, we reason 300-400 USD a month could be earned, 

• Ad Revenue - Pressum Casual will be free to access, however, users will be shown ads throughout their experience. We aim to make these ads minimally intrusive, instead focusing on improving click through rate for higher revenues. We could observe user behavior and suggest relevant products like journal subscriptions over more generic products. This will lead to a better user experience while preserving revenue. 

• Sale of Subscriptions - However, a small user base will still be willing to pay a premium for no ads. We could charge 400 INR (around 5 USD) for a monthly subscription to do this. We also feel a more expensive 800 INR (around 10 USD) monthly subscription for early access could be a potential hit among enthusiasts or teachers looking to provide their classes with a unique experience.