Inspiration

Our project was largely inspired by the pervasive issue of fake news on social media and its detrimental impact on public discourse and trust. Observing how misinformation spreads rapidly each day, we recognized the urgent need for tools that can safeguard the digital ecosystem by ensuring only credible content receives visibility and support. Additionally, our daily interactions with ads from various brands across social media and the internet highlighted the intersection of advertising and information dissemination. This combination spurred our motivation to develop a solution that not only combats fake news but also ensures that advertising dollars are wisely invested in webpages that are reliable and aligned with brand values.

What it does

Our tool acts as a critical gatekeeper for digital advertisements by vetting and verifying the legitimacy of content before ads are placed. By integrating advanced algorithms with real-time data analysis, the tool evaluates the authenticity of news articles and social media posts, ensuring advertisers support and foster a truth-based digital environment. This system not only protects brands but also helps in cultivating a well-informed public by reducing the reach of false information.

How We Built It

Our project integrates various complex components to create a robust tool for combating misinformation. Here’s a detailed breakdown of our development process:

Data Collection: Utilized web crawlers and APIs to gather a continuous stream of news articles and social media posts, ensuring a diverse data source encompassing multiple perspectives.
Preprocessing: Rigorous text cleaning including the removal of harmful content, neutralizing tone, and normalizing text. These tasks standardized input data for more accurate analysis.
Gemini API Integration: Integrated the Gemini-1.5-pro-latest API to analyze text for misinformation signals by identifying language patterns and factual grounding. Each piece of content was assigned a "fake news score" indicating its likelihood of being misinformation.
Ad Network Integration: Built to interface seamlessly with ad networks, preventing ads from being displayed on flagged articles and better targeting ads based on factual content.
Action & Reporting Mechanisms: Flagged articles could be displayed with a warning label, downranked in search results, or reported back to the originating platform.
Criteria for Fact-Checking: Established robust criteria that included evaluating source credibility, verifying accuracy, assessing logical consistency, examining objectivity, and scrutinizing language use.
Technology Stack: Developed using Python with Django for server implementation. HTML/CSS built the frontend and MySQL stored data via Django ORM.
Dataset: Created from legitimate articles and fake news sources, rich in political and international news, to train the tool in recognizing misinformation patterns.

Challenges we ran into

We encountered several challenges such as the StopCandidateException, where the process halted when explicitly harmful content was detected without a clear standard for flagging text. Keeping the tool updated with the latest news and the limited language support from the Gemini 1.5 pro API were significant hurdles that impacted global usability.

What we learned

The development deepened our expertise in API integration, data analysis, and digital advertising dynamics. We gained valuable insights into the complexities of discerning misinformation across various contexts and languages.

Accomplishments that we're proud of

According to data from WordStream, the global digital advertising and marketing market is estimated at $667 billion for 2024 and is projected to reach $786.2 billion by 2026. In this rapidly growing field, we take immense pride in having developed a tool that significantly impacts advertisement allocation and enhances the integrity of digital advertising. Our solution effectively combats misinformation, helping brands maintain their reputation and contributing positively to the internet community.

What's next for Legitimacy of Web Content

As we look to the future, we aim to expand our prototype into full-scale deployment, fostering partnerships with major advertising platforms, including programmatic agencies. We plan to enhance the tool's language capabilities and refine our AI algorithms for more precise detection. These improvements will enable broader adoption and more effective mitigation of misinformation, paving the way for more authentic and reliable digital advertising environments.