Fraud Detection via Chatbot Stalling in Voice Calls
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The widespread sharing of personal information due to advancements in technology has led to security attacks, where fraudsters easily access and manipulate customer account information, resulting in a high likelihood of fraudulent calls attempting to change contact details for account takeover.
Innovation Solution
A machine learning-based system that identifies potentially fraudulent calls by analyzing call transcripts and transferring such calls to a chatbot, which stalls the caller and prevents future fraudulent attempts by wasting their time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If personal information is widely shared and stored on technology platforms, then technology services can be provided to users, but security attacks and fraud increase as a result
Solution Approach 1:
The patent introduces an intermediary system between the stored personal information and potential fraudsters. This system uses machine learning models to analyze call transcripts and identify fraudulent calls, acting as a mediator that blocks harmful access while allowing legitimate services to continue. The intermediary detects patterns in communication to distinguish between valid customer service requests and fraudulent attempts.
Solution Approach 2:
The patent implements preliminary action by proactively identifying and stalling fraudulent calls before they can successfully compromise account information. The system analyzes call transcripts in real-time or near-real-time, detects fraudulent patterns, and intervenes by transferring calls to chatbots that waste the fraudster's time, preventing the harmful outcome before it occurs.
2Ease of operation
If fraudsters can easily access personal information through phone calls, then account takeover becomes simple, but implementing security measures increases system complexity
Solution Approach 1:
The patent implements self-service by enabling the security system to automatically analyze and classify calls without requiring manual human review. The machine learning models autonomously process call transcripts, identify fraudulent patterns, and route calls appropriately. This automation reduces the complexity burden on operators while maintaining effective security monitoring.
Solution Approach 2:
The patent changes the parameter of call analysis from manual human judgment to automated machine learning classification. By transforming the security detection process into an automated parameter-based system that analyzes transcript features, call duration, and communication patterns, the system manages complexity through standardized algorithms rather than manual procedures.
3Reliability
If a chatbot is used to stall fraudulent calls, then fraud prevention effectiveness increases, but call handling time increases for legitimate customers
Solution Approach 1:
The patent applies partial action by using chatbots to stall only fraudulent calls, not all calls. The machine learning system selectively identifies fraudulent calls and routes them to chatbots for stalling, while legitimate calls continue to be handled by human agents or automated systems without unnecessary delays. This partial application of the stalling mechanism prevents widespread time loss while maintaining fraud prevention effectiveness.
Solution Approach 2:
The patent implements feedback by continuously monitoring call outcomes and refining the machine learning model's ability to distinguish fraudulent from legitimate calls. As the system accumulates more data about fraud patterns and legitimate customer behavior, it improves its classification accuracy, reducing false positives that would incorrectly route legitimate customers to chatbots and cause unnecessary delays.
Data Source
AI summary
A server can receive a plurality of records at a databases such that each record is associated with a phone call and includes at least one request generated based on a transcript of the phone call. The server can generate a training dataset based on the plurality of records. The server can further train a binary classification model using the training dataset. Next, the server can receive a live transcript of a phone call in progress. The server can generate at least one live request based on the live transcript using a natural language processing module of the server. The server can provide the at least one live request to the binary classification model as input to generate a prediction. Lastly, the server can transmit the prediction to an entity receiving the phone call in progress. The prediction can cause a transfer of the call to a chatbot.