Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Cognitive Crowd Vibe Analyzer

IP.com Disclosure Number: IPCOM000250230D
Publication Date: 2017-Jun-14
Document File: 5 page(s) / 296K

Publishing Venue

The IP.com Prior Art Database

Abstract

Using a multitude of available sources, aggregate their information, estimate the vibe of the crowd and recommend dispatch actions. Patterns, especially suspicious patterns, will be detected trough real-time and decentralized scanning and evaluating of different data sources. The detected pattern are compared with previous occurred patterns based on similar characteristic and a list of recommended actions. The final analysis results of the suspicious pattern are sent to a control center.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 31% of the total text.

TITLE: Cognitive Crowd Vibe Analyzer

Current surveillance is mostly done using surveillance cameras, partially through social insight and patrolling to scan the situation within a particular area. However, most of the monitoring activities are run manually and the assessment of the atmosphere relies on the experience of the authorities.

A limited set of data sources which are not aggregated to allow meaningful insight analysis for appropriate situation monitoring results in delaying appropriate dispatch actions.

Another disadvantage is the frugality of the tools which doesn’t include any kind of learning mechanism to improve the established surveillance infrastructure based on past events.

The use of a multitude of available sources, aggregate their information, estimate the vibe of the crowd and recommend dispatch actions is shown in figure below.

Figure 1: Cognitive crowd vibe analyzer overview

Patterns, especially suspicious patterns, will be detected trough real-time and decentralized scanning and evaluating of different data sources like live surveillance footage (e.g. unusual number of people moving fast), additional sensor data (e.g. high carbon dioxide level within closed environments, voice/noise and speech recognition and analysis), live social media feeds (text analytics for emotion, sentiment and key words patterns), mobile phone position tracking, traffic navigation systems and other data sources.

Once one data source listener detects a (suspicious) pattern the extraction of additional data from the mentioned data source is triggered.

The gathered data is aggregated, scored and compared against given (suspicious) patterns (combination of data values from various data sources and across data types) and the occurring (suspicious) pattern confirmed and codified for comparison operations.

Once confirmed the detected pattern is compared with previous occurred patterns based on similar characteristic and a list of recommended actions and other additional information is attached to the detected pattern data. Simultaneously social media feeds listeners are activated using dynamically defined listening criteria (e.g. combination of hash-tags/geo-location) specific for the detected pattern to evaluate and isolate the impacted area.

The final analysis results of the (suspicious) pattern are sent to a ‘multi-device’-enabled control center and displayed to a control center agent.

The component model shows how data from multiple sources is combined and processed in the system 100.

Figure 2: Components of the crowd vibe analyzer

Data from various data sources 110-113 covering various locations collected in one localized person database 130 and one localized flux database 131. Collecting data from various locations has the advantage that several events (hot spots) can commonly analyzed, which is especially useful when crowds are merging or diverting or common vibe patterns can be observed.

Individual persons can be identified by surveillance cam...