Automated scoring of student essays is fast, accurate, and affordable. That was the conclusion drawn from two prize competitions sponsored by the William and Flora Hewlett Foundation. ASAP began in February of 2012 with a demonstration of capabilities of the eight largest testing vendors. The “bake off” was hosted on the Kaggle platform and, as Mark Shermis and Ben Hamner reported, demonstrated that current scoring engines could match expert graders across eight sets of essays. A case study, “Automated Student Assessment Prize Phase One and Phase Two: A Case Study to Promote Focused Innovation in Student Writing Assessment,” was published in January.
Perhaps the biggest surprise of the ASAP competitions was the stunning performance of LightSide, an open scoring engine developed at Carnegie Mellon University. Grad student Elijah Mayfield and the open source code held their own against testing companies and data scientists from around the world.
Mayfield left CMU and formed a company around the scoring engine and today is rolling out a process for schools to join a spring pilot project. Some detail about the pilot can be found at the LightSide application page and in a press release below.
Pittsburgh’s latest education company is giving local schools the first chance to see how young writers can benefit from advanced machine learning technology.
The new platform comes from LightSide, a tech spin-out of Carnegie Mellon University that improves student writing skills by using machine learning to assess writing. When deployed, their system automatically gives feedback on student essays, encourages revision, and eases the burden of grading for teachers struggling to maintain high standards for classroom writing.
With this new platform, students write through an online interface, either on a computer or tablet. Once a first draft is complete, students will get instant, automated feedback on their progress, highlighting strengths and weaknesses of their writing. This guidance will continue for each draft, at the student’s own pace, until the final version is submitted to their teacher. Suggested scores are instantly given to teachers, who then review, add feedback, and assign a final grade to the student’s work.
Chartiers Valley School District, located 6 miles southwest of downtown Pittsburgh, has already signed on as the first participant. “We’ve met with our English Language Arts teachers, and they’re thrilled to be involved in this development,” said Yvonne Hawkins, the district’s Assistant Superintendent for Curriculum. Up to 10 districts will be selected through the application process, which is open through November 24.
The platform, aimed at middle and high school students, has been under development since the company’s founding in March 2013. The underlying machine learning algorithms were developed at Carnegie Mellon University, while LightSide’s founders were Ph.D. students at the Language Technologies Institute. Alumni of the LTI program have been leaders on the teams that built Google’s search engine, Apple’s Siri personal assistant, IBM’s Watson platform for question-answering, and many other computer programs that process human language.
Earlier in 2013, the Bill & Melinda Gates Foundation named LightSide as one of only 29 national winners of its “Literacy Courseware Challenge.” The foundation sought new ideas that could help 4th to 8th-grade students master the reading and writing skills required by the new Common Core State Standards.
– CONTACT –
All press inquiries, including setting up interviews with founder Elijah Mayfield, should be sent to: [email protected]
– MORE INFORMATION –
http://www.lightsidelabs.com/pilot for applications to participate in the pilot. http://www.lightsidelabs.com for more about the company’s technology and team. http://cbsn.ws/QIsU0Z for information about 2012’s record-low SAT essay scores. http://www.cvsd.net to learn about Chartiers Valley School District. http://bit.ly/YzK07o for more information about the Literacy Courseware Challenge. http://www.cmu.edu/olympus/ to learn about Project Olympus and other CMU startups.