Personalized Ad Generation via NLP Intent Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital advertisement targeting technologies rely on inferences and are inefficient in placing ads in front of the right prospects at the right time, due to challenges in processing large data sets, high costs, and sensitivity to private data, leading to reduced effectiveness and higher costs.
Innovation Solution
A system that generates personalized advertisements based on digital data from users' media posts, using natural language processing and tagging models to identify purchase intent and attributes, and publishes them in real-time when certain thresholds are met, reducing reliance on sensitive data and lowering costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current ad targeting technologies use demographic, behavior, and social networking data to identify prospects, then ads can be targeted to potential customers, but the targeting is still based on inferences and guesses rather than actual purchase intent
Solution Approach 1:
The patent uses social media posts as an intermediary signal to detect actual purchase intent. Instead of directly inferring intent from demographic data, the system monitors user-generated content (posts, comments, shares) that explicitly indicate purchase intentions, using these posts as a mediator between the user and the ad delivery system.
Solution Approach 2:
The patent replaces the mechanical inference-based targeting system with an information-based system that processes natural language from social media posts. Natural language processing and machine learning models analyze text content to detect purchase intent signals, substituting statistical inference with semantic understanding.
2Measurement precision
If advertisers process large sets of data to improve targeting, then ad relevance may improve, but costs increase and processing complexity increases
Solution Approach 1:
The patent extracts only the most relevant information from social media posts - specifically text content that indicates purchase intent. Instead of processing entire datasets including images, videos, and metadata, the system focuses on extracting and analyzing only the textual elements that contain purchase signals, significantly reducing processing complexity.
Solution Approach 2:
The patent segments the advertising system into distinct modules: social media data collection, text extraction, natural language processing, purchase intent detection, and ad delivery. This segmentation allows each component to handle specific tasks independently, reducing overall system complexity and enabling parallel processing.
3Measurement precision
If advertisers use more private or sensitive data to improve targeting, then ad effectiveness may increase, but user privacy concerns increase and data sensitivity increases
Solution Approach 1:
The patent uses publicly available social media posts that users voluntarily share. The purchase intent information is self-disclosed by users through their own posts and interactions, eliminating the need for advertisers to collect or infer sensitive personal data. Users effectively provide the data themselves through their public communications.
Solution Approach 2:
The patent uses social media posts that serve multiple purposes: they are public communications for social interaction, and simultaneously serve as purchase intent signals for advertising targeting. This multi-functionality allows the system to use existing user-generated content without requiring additional data collection or compromising privacy.
4Loss of time
If advertisements are placed based on inferences about purchase likelihood, then ads can be targeted in advance, but ads are not placed at the right time when purchase decision is manifested
Solution Approach 1:
The patent continuously monitors social media posts in real-time, maintaining a ready state to detect purchase intent signals as they emerge. The system has pre-configured detection capabilities and can immediately act on detected signals, combining preliminary preparation with real-time responsiveness.
Solution Approach 2:
The patent implements a feedback loop where social media posts are continuously monitored, purchase intent is detected through natural language processing, and ad delivery is adjusted in real-time based on detected intent. This closed-loop system ensures ads are delivered at the precise moment purchase intent is expressed, with continuous refinement of detection accuracy.
Data Source
AI summary
In an aspect of the disclosure, a method, a computer-readable medium, and an apparatus are provided. The apparatus may receive digital data associated with digital media communications. The apparatus may input the textual data into a NLP model. The apparatus may obtain intent data associated with an item as an output of the NLP model. The apparatus may input the intent data into a tagging model. The apparatus may obtain an attribute tag as an output of the tagging model. The apparatus may determine whether the intent data meets a likelihood threshold. The apparatus may determine whether the attribute tag meet a relevance threshold. The apparatus may output the intent data and the attribute tag to an external device upon determining that the intent data meets the likelihood threshold and that the attribute tag meets the relevance threshold.