Automated Content Verification Solution for Prominent News and Media House
About
The client is a prominent and largest news publishing house, it has its pillars rooted in the media and entertainment sector of India. Apart from print media, it has a reputed presence in Medias like FM radio’s the women’s magazines, a financial daily, English film magazine, entertainment channel and most importantly the active News channel of India.
Project Highlights
The Brainvire team successfully crafted a .Net MVC application that will manage Advertisements submission to news publishing house’s varied editions. With the integration of eUnagi services the posted content will be checked for profanity and accordingly will be updated with vital information so that it can become searchable. eUnagi services will perform automated checks on hundreds of submissions and if any profanity is detected in images then those will be deleted and on remaining images services like Face Recognition, Emotion Recognition, Place Recognition, etc will be implemented. eUnagi profanity check system has the capability to quickly and accurately detect words and phrases that are considered profane or offensive. The system allowed users to customize their own profanity lists and set up filters that are specific to their needs.
The system provides multiple filtering options allowing users to adjust the level of profanity detection. For example, some users may want to detect profanity in texts, while others may want to detect it in audio or video. The system is able to understand the context of a word or phrase to determine whether it is offensive. For example, some words may be acceptable in certain contexts, but not in others. The system provides users with detailed reports of any profanity it detects. This allows users to quickly identify which words need to be removed or modified.
The eUnagi screening technology can detect objectionable words and phrases quickly. The technology allowed users to create their own profanity lists and filters. For example, some users may like to detect swearing in audio or video. The system may judge offensiveness depending on context. The technology generates detailed reports on profanity. This helps users identify phrases to delete or modify.
The Challenges
- Creating A Profanity Filter:It was challenging for the client to detect the offensive language in every given post format, whether it be text, photos, audio, or videos, and hence they needed an advanced solution.
- Quick Geolocation Detection:The client manages a large amount of data, and it is time-consuming to analyze the geolocation of an image's coordinates manually.
- The Use Of Facial Recognition:The client desired several different types of services, including recognition of facial expressions and features, identification of locations and objects, and object detection.
- Cloud Computation:The most significant challenge encountered was distributing the Cloud computing resources.
Tech Stack
jQuery
Microsoft SQL Server
Python
.Net core
Azure cloud
Microsoft Cognitive Services
C# Microsoft .Net
Result
Offensive Intent Detection
Using the Levenshtein automata, the library recognizes exact profane word matches and derivative and distorted profane terms while avoiding dictionary words containing profane words. Signal potentially hazardous content as suspicious or malevolent language, and maintain content quality efficiently.Improved Security
The application was able to determine the geolocation of an image. This geolocation data added a degree of security to a system by enabling it to detect when a picture was captured in an unexpected area, which could signify a security compromise.Sorting with Facial recognition
This helped ensure that the process was error-free, as it reduced the number of mistakes humans could have brought on. This helped adopt a speedy course of action.Workload Automation
The client managed the required workload better, and our team proposed automation through Cloud computing.
