Move Over, Siri! University of Southern California Researchers Develop Improv-Based Chatbot

What would conversations with Alexa be like if she was a regular at The Second City?

Jonathan May, research lead at the USC Information Sciences Institute (ISI) and research assistant professor of computer science at USC's Viterbi School of Engineering, is exploring this question with Justin Cho, an ISI programmer analyst and prospective USC Viterbi Ph.D. student, through their Selected Pairs Of Learnable ImprovisatioN (SPOLIN) project. Their research incorporates improv dialogues into chatbots to produce more engaging interactions.

The SPOLIN research collection is made up of over 68,000 English dialogue pairs, or conversational dialogues of a prompt and subsequent response. These pairs model yes-and dialogues, a foundational principle in improvisation that encourages more grounded and relatable conversations. After gathering the data, Cho and May built SpolinBot, an improv agent programmed with the first yes-and research collection large enough to train a chatbot.

The project research paper, "Grounding Conversations with Improvised Dialogues," was presented on July 6 at the Association of Computational Linguistics conference, held July 5-10.

Finding Common Ground

May was looking for new research ideas in his work. His love for language analysis had led him to work on Natural Language Processing (NLP) projects, and he began searching for more interesting forms of data he could work with.

"I'd done some improv in college and pined for those days," he said. "Then a friend who was in my college improv troupe suggested that it would be handy to have a 'yes-and' bot to practice with, and that gave me the inspiration--it wouldn't just be fun to make a bot that can improvise, it would be practical!"

The deeper May explored this idea, the more valid he found it to be. Yes-and is a pillar of improvisation that prompts a participant to accept the reality that another participant says ("yes") and then build on that reality by providing additional information ("and"). This technique is key in establishing a common ground in interaction. As May put it, "Yes-and is the improv community's way of saying 'grounding.'"

Yes-ands are important because they help participants build a reality together. In movie scripts, for example, maybe 10-11% of the lines can be considered yes-ands, whereas in improv, at least 25% of the lines are yes-ands. This is because, unlike movies, which have settings and characters that are already established for audiences, improvisers act without scene, props, or any objective reality.

"Because improv scenes are built from almost no established reality, dialogue taking place in improv actively tries to reach mutual assumptions and understanding," said Cho. "This makes dialogue in improv more interesting than most ordinary dialogue, which usually takes place with many assumptions already in place (from common sense, visual signals, etc.)."

But finding a source to extract improv dialogue from was a challenge. Initially, May and Cho examined typical dialogue sets such as movie scripts and subtitle collections, but those sources didn't contain enough yes-ands to mine. Moreover, it can be difficult to find recorded, let alone transcribed, improv.

The Friendly Neighborhood Improv Bot

Before visiting USC as an exchange student in Fall 2018, Cho reached out to May, inquiring about NLP research projects that he could participate in. Once Cho came to USC, he learned about the improv project that May had in mind.

"I was interested in how it touched on a niche that I wasn't familiar with, and I was especially intrigued that there was little to no prior work in this area," Cho said. "I was hooked when Jon said that our project will be answering a question that hasn't even been asked yet: the question of how modeling grounding in improv through the yes-and act can contribute to improving dialogue systems."

Cho investigated multiple approaches to gathering improv data. He finally came across Spontaneanation, an improv podcast hosted by prolific actor and comedian Paul F. Tompkins that ran from 2015 to 2019.

With its open-topic episodes, about a good 30 minutes of continuous improvisation, high quality recordings, and substantial size, Spontaneanation was the perfect source to mine yes-ands from for the project. The duo fed their Spontaneanation data into a program, and SpolinBot was born.

"One of the cool parts of the project is that we figured out a way to just use improv," May explained. "Spontaneanation was a great resource for us, but is fairly small as data sets go; we only got about 10,000 yes-ands from it. But we used those yes-ands to build a classifier (program) that can look at new lines of dialogue and determine whether they're yes-ands."

Working with improv dialogues first helped the researchers find yes-ands from other sources as well, as most of the SPOLIN data comes from movie scripts and subtitles. "Ultimately, the SPOLIN corpus contains more than five times as many yes-ands from non-improv sources than from improv, but we only were able to get those yes-ands by starting with improv," May said.

SpolinBot has a few controls that can refine its responses, taking them from safe and boring to funny and wacky, and also generates five response options that users can choose from to continue the conversation.

SpolinBot #Goals

The duo has a lot of plans for SpolinBot, along with extending its conversational abilities beyond yes-ands. "We want to explore other factors that make improv interesting, such as character-building, scene-building, 'if this (usually an interesting anomaly) is true, what else is also true?,' and call-backs (referring to objects/events mentioned in previous dialogue turns)," Cho said. "We have a long way to go, and that makes me more excited for what I can explore throughout my PhD and beyond."

May echoed Cho's sentiments. "Ultimately, we want to build a good conversational partner and a good creative partner," he said, noting that even in improv, yes-ands only mark the beginning of a conversation. "Today's bots, SpolinBot included, aren't great at keeping the thread of the conversation going. There should be a sense that both participants aren't just establishing a reality, but are also experiencing that reality together."

That latter point is key, because, as May explained, a good partner should be an equal, not subservient in the way that Alexa and Siri are. "I'd like my partner to be making decisions and brainstorming along with me," he said. "We should ultimately be able to reap the benefits of teamwork and cooperation that humans have long benefited from by working together. And the virtual partner has the added benefit of being much better and faster at math than me, and not actually needing to eat!"

About the USC Viterbi School of Engineering

Engineering began at the University of Southern California in 1905. Nearly a century later, in 2004, the school was named the USC Viterbi School of Engineering, honoring alumnus Andrew J. Viterbi, inventor of the Viterbi algorithm, a key to cell phone technology and numerous data applications. The school is consistently ranked among the top engineering programs in the world. It enrolls more than 7,000 undergraduate and graduate students, taught by 188 tenured and tenure-track faculty, with more than 90 endowed chairs and professorships. The school offers comprehensive programs in all major areas of engineering, from aerospace and astronautical engineering, to biomedical, chemical, civil and environmental engineering, to computer science and materials science, and to electrical engineering, industrial and systems engineering and mechanical engineering. With annual research expenditures exceeding $207M, the school is home to the Information Science Institute and more than 45 other research centers. Its more than 50,000 alumni span the globe. USC Viterbi is also home to the pioneering distance learning program DEN@Viterbi, ranked as the top such program in the nation the last two years. USC Viterbi is the only engineering school in the nation with three top-10 USNWR rankings: Best Graduate Schools, Best Online Graduate Engineering Programs, and Best Online Information Technology Programs. 

Snapchat Rolls Out ‘Here For You’ Features in the Indian Market

Snap Inc. (NYSE: SNAP) today, launched Here for You in India, a feature that provides in-app support to Snapchatters who may be experiencing a mental health or emotional crisis or those who want to learn more about these issues in order to help friends. Mariwala Health Initiative and the Manas Foundation are the first partners in India, who have created resources around anxiety, depression, loneliness, suicide and general mental health.

The creation of Here for You was informed by research conducted late last year into how Snapchatters and their friends experience mental health. Snapchat found an overwhelming majority of Snapchatters experience feelings of stress and anxiety and that their friends are the first people they turn to when they need help, more than professionals - or even their parents. They also see their friends as the most positive force in their lives - consistent with other studies that have shown that spending time with friends can be one of the most effective ways to combat feelings of loneliness and depression. The research also revealed that many young people are deeply interested in understanding these issues, and how they can support friends who are struggling with them. 

“At Snap, we have always been inspired by the power real friendships have in supporting health and happiness. Now more than ever, in a socially distanced world, we see friendship and connection is vital to our wellbeing. Here For You was designed to empower our community with resources to help themselves, as well as educate on how to spot signs in loved ones. It’s great to partner with Mariwala Health Initiative and the Manas Foundation to do this - organisations with a great wealth of experience and expertise in India,” said Jennifer Park Stout, VP of Global Public Policy, Snap.

“These are unprecedented times. COVID-19 has changed how we live our lives, how we work or study, how we connect with each other and more. We are collectively experiencing feelings of worry, stress, tension, anxiety, numbness, sadness - which are all routine responses to periods of crises. It is therefore crucial to have mental health resources available and accessible for those experiencing distress. Platforms like Snapchat provide a powerful medium to disseminate these resources. Our videos discuss how to deal with stress and anxiety, how to support a loved one experiencing distress and ways to enhance mental well-being through mindfulness/grounding exercises,” said Raj Mariwala, Director, Mariwala Health Initiative.

“For over two decades we have been striving to demystify mental health in India. Students, facing school and college closures, are missing out on robust dialogues, outdoors, experiential learning and interpersonal interactions that add to their well-being. For many, the loss of routine is also leading to erratic eating and sleeping schedules; which is directly linked to one’s mental health. This launch is a unique opportunity for users to relate with mental health as part of their everyday life. We want to tell them that it is ok to talk about mental health and Snapchat and Manas are here for you,” said Naveen Kumar

Trustee & Psychologist, Manas Foundation.

Here For You will be available to Snapchatters in India from today. They can access resources by searching for words such as ‘anxiety’, ‘depression’, ‘loneliness’, ‘suicide’, ‘mental health’ and ‘wellbeing’. The content covers topics such as ways to cope with mental health issues and how to spot signs in loved ones, as well as questions from the community, answered by clinical psychologists and mental health professionals. Each ends with further information on the Manas Foundation and Mariwala Health Initiative, and advice on how to contact a trained counsellor directly for help.

The Here For You feature  is part of a concerted effort designed to educate and empower the Snapchat community to look after themselves, and support friends who might be struggling with their own social and emotional wellbeing. Other features announced recently at Snap’s Partner Summit in June include a partnership with Headspace, new Snap Originals and expanded in-app tools.

Ex-lovers most popular online search on the Net

Most people look for their long lost love on the internet, a new survey has found.

The study found that one out of four people are using social networking sites like Facebook to search a childhood sweetheart.

Search engine Ask Jeeves claims that 9% even confessed trying to know details of a one-night stand.

Nearly 37% said that they wanted to learn about their ex to just “see what they were doing these days”, reports the British tabloid the Sun.

Surprisingly, 4% even looked for former flames just to inform them how happy they were without them, while 3% searched to find out how miserable their ex were.

Apart from searching for previous lovers the poll also suggested that a trend of “vanity searching”, which means looking for yourself on the net, has become increasingly popular.

Agencies

Why Amazon, Microsoft, Yahoo! oppose Google book project?

Amazon, Microsoft and Yahoo! are planning to join a coalition of library associations and non-profit groups in opposing Google's ambitious book scanning project, US newspapers reported on Friday.

The New York Times and Wall Street Journal said the technology heavyweights have agreed to form what is tentatively being called the Open Book Alliance to challenge Google's class action settlement with authors and publishers.

The settlement, which gives the Internet search and advertising giant the rights to commercialize digital copies of millions of books, is already facing anti-trust scrutiny from the Justice Department and awaiting court approval.

Gary Reback, an anti-trust lawyer in Silicon Valley who is acting as counsel to the Alliance, told the Times the book deal "has enormous, far-reaching anticompetitive consequences that people are just beginning to wake up to."

Reback, who helped persuade the Justice Department to file its anti-trust case against Microsoft in the 1990s, said the group includes the Internet Archive, a San Francisco non-profit which maintains a digital library of websites.

The Times said the group plans to make a case to the Justice Department that the arrangement is anticompetitive.

It said that members of the Alliance were likely to file objections independently with the US District Court in New York which is set to hold a "fairness hearing" on the deal on October 7.

Microsoft and Yahoo! confirmed to the Times that they were participants while Amazon refused to comment.

Peter Brantley, a director of the Internet Archive, said the Special Libraries Association, the New York Library Association and the American Society of Journalists and Authors were planning to join the group.
He told the Journal its membership would be formally disclosed in the next couple
of weeks.

Brantley said members of the coalition all see problems with the settlement and are pushing for revisions, but not all necessarily want to see it blocked.

The Google Book Search project has come under fire from a number of quarters, including from groups worried about privacy.

The American Civil Liberties Union, the Electronic Frontier Foundation and the Samuelson Law, Technology and Public Policy Clinic of the University of California at Berkeley recently wrote to Google chief executive Eric Schmidt expressing concerns about privacy aspects of the deal.

"Given the long and troubling history of government and third party efforts to compel libraries and booksellers to turn over records about readers, it is essential that Google Books incorporate strong privacy protections in both the architecture and policies of Google Book Search," they said.

Google reached a settlement last year with the Authors Guild and Association of American Publishers on a copyright infringement lawsuit they filed in 2005 over Google's plan to scan millions of books and put them online.

Under the settlement, Google agreed to establish an independent "Book Rights Registry," which will provide revenue from sales and advertising to authors and publishers who agree to digitize their books.

Microsoft shut down its own book scanning project in May of last year.

Agencies

Did an Indian surrender domain name to Google?

Internet search giant Google has won a cybersquatting case at the World Intellectual Property Organisation (WIPO) against an Indian who had tried to block the domain name 'googblog.com'.

According to the information available with the WIPO, Geneva-based WIPO Arbitration and Mediation Center has ordered the transfer of domain name to the US-based search giant after Herit Shah of Gujarat offered to surrender the disputed name to Google.

Google had challenged the registering of domain name 'googblog.com' by Shah at WIPO stating that it was confusingly similar to its trademark on which the company has rights.

Cybersquatting is an illegal activity of buying and officially recording an address on the internet that is the name of an existing company or a well-known person, with the intention of selling it to the owner in order to make money.

As per the information available with the WIPO, Google filed the complaint against Shah on March 26 this year. However, the disputed name has been registered by Shah since September 25, 2008.

WIPO is a specialised agency of the United Nations for developing a balanced and accessible international system in the field of intellectual property rights.

The California-headquartered firm has been using the name 'GOOG' as a NASDAQ financial stock ticker since 2004. The company has used the trademark GOOGLE since the inception of its business in 1997.

The search giant operates a blog service under the brand 'Blogger'.

As per the details available with WIPO, a pre-complaint correspondence between the parties (Google and Shah) failed to resolve the dispute.

However on May 2, after commencement of administrative proceedings, Shah stated before the panel that the registration of domain name was in bad faith and was an infringement of intellectual property.

"I was in a bad faith that I can legally keep the domain googblog.com ... I really did very unfair to Google. I sincerely apologise to Google for infringement, misuse of their intellectual property (GOOGBLOG.COM)," Shah stated.

The WIPO panel found in this case the consent-to-transfer request replaces the need to assess the matter under the elements of its Uniform Domain Name Dispute Resolution Policy and ordered the transfer of the domain name to Google.

Agencies

Amazon.com finds competitor in Google in e-books

Seeking to find common ground with authors, who have complained about copyright violations through search services, Google plans
later this year to begin distributing and selling e-books on behalf of its publishing partners.

"We've consistently maintained that we're committed to helping our partners find more ways to make their books accessible and available for purchase," Google spokesman Gabriel Stricker said confirming the move first reported by the New York Times.

"By end of this year, we hope to give publisher partners an additional way to sell their books by allowing users to purchase access to partner programme books online," he said in an e-mail to InformationWeek, a leading source for information technology news.

"We want to build and support a digital book ecosystem to allow our partner publishers to make their books available for purchase from any Web-enabled device."

Google is anxious to find common ground with authors, who have complained about copyright violations in the past through services like Google Book Search.

Formerly known as Google Print, Google Book Search was introduced in 2004 and targeted by publishers and their lawyers the following year for digitising books without the permission of copyright holders.

A proposed settlement of that lawsuit is currently being reviewed by the courts and the US Department of Justice.

Google's e-book sales service will be made available to participants in the Google Book Service Partner Programme, a marketing programme for promoting books through Google Book Search.

If Google succeeds in making peace with authors and publishers, it may find itself competing more directly against Amazon.com, the publication said.

Amazon gave up competing against Google Search in 2006 when it closed its A9 search engine, but Amazon Web Services, the company's on-demand computing infrastructure service, remains a strong contender against Google App Engine.

With its Kindle e-book reading devices, Amazon has been building the infrastructure and market for electronic texts on portable devices, a transition in reading technology that's been anticipated for a decade, but never fully realised.

Apple too will soon release its iPhone 3.0 operating system, which will bring e-book sales opportunities to the iPhone's many e-book reading apps, InformationWeek said.

Apple is rumoured to be working on tablet computing device, a form factor ideal for reading e-books. Other players, like Sony, see a future in e-books, too.

Agencies