Principal Investigator

Aaron Steinfeld

Status

Completed

Start Date

Jan. 1, 2014

End Date

Dec. 31, 2014

Project Type

Research Advanced

Grant Program

MAP-21 TSET - Tier 1 (2012 - 2016)

Grant Cycle

2014 TSET UTC

Visibility

Public

Abstract

The primary source of information for rider safety with respect to dynamic events such as cancelled buses and snow detours is the transit system website and, often, the agency's Twitter feed. However, users require a lot of effort to understand these messages, since at any point in time, a transit system has dozens of problems. A user must read through all the messages to understand if any are relevant. Finally, it is difficult for riders to forecast how this will impact the fullness of alternative routes. 
Our project goal is to create a natural language processing system that automatically extracts relevant information from each message. The extracted information can be used in multiple ways to improve the interactive experience of the user and keep them better informed. For example, messages could be surfaced to the user in an interface if it addresses a stop or route that the user typically uses, or for which the user is currently requesting data. This, combined with vehicle fullness data from the Tiramisu system, will allow riders to identify alternate transit trip options that are not too full to board. 
We will start with agency Twitter feeds as an input. While not all transit agencies use Twitter, many cities rely on this content channel to transmit real-time safety, detour, and status updates. Processing this information source is not always and easy problem. There is often an inexact match between the official entity names and tweet content. We will also work on the difficult problem of event extraction, that is, attaching time duration to the events extracted. 
Processing and distributing information about abnormal transit operations has a direct impact on rider safety and efficient travel. Unnecessary exposure to inclement weather while waiting for a bus that will never arrive can have a significant negative health impact.    

Description

-

Timeline

-

Strategic Description / RD&T

Deployment Plan

-

Expected Outcomes/Impacts

-

Expected Outputs

TRID

Individuals Involved

Email	Name	Affiliation	Role	Position
steinfeld@cmu.edu	Steinfeld, Aaron	Robotics Institute	PI	Faculty - Tenured

Budget

Amount of UTC Funds Awarded

$50000.00

Total Project Budget (from all funding sources)

$50000.00

Documents

Type	Name	Uploaded
Final Report	tiramisu__information_from_live_data_streams.pdf	April 2, 2018, 5:09 a.m.

Match Sources

No match sources!

Partners

No partners!