World Health Organization: WHO launches new search feature for questions on COVID-19. “WHO’s COVID-19 webpage now features an enhanced natural language processing search bar, which understands questions posed in everyday language and more accurately delivers answers to those queries. Unlike traditional index-based search that delivers links, when someone enters a question about COVID-19 into the search bar on WHO’s COVID-19 page, the new search finds the most accurate information related to that question from WHO’s website. Yext, the technology company that developed this search function, provides WHO’s web team with regular feedback on the questions that visitors are asking so that WHO can adapt information on the web to meet the demand.”
Neowin: TextFooler tricks venerable NLP models like BERT into making wrong predictions. “Jargon aside, the program swaps the most important words with synonyms within a given input to modify how the models interpret the sentence as a whole. While these synonyms might be common to us and the sentence would still have similar semantics, they made the targeted models interpret the sentences differently.”
Search Engine Journal: Google’s BERT Rolls Out Worldwide. “Google announced via Twitter that their BERT algorithm is now rolling out worldwide. BERT will enable Google to better understand search queries.”
IEEE Spectrum: Natural Language Processing Dates Back to Kabbalist Mystics. “While specific technologies have changed over time, the basic idea of treating language as a material that can be artificially manipulated by rule-based systems has been pursued by many people in many cultures and for many different reasons. These historical experiments reveal the promise and perils of attempting to simulate human language in non-human ways—and they hold lessons for today’s practitioners of cutting-edge NLP techniques. The story begins in medieval Spain.”
Search Engine Land: Why you may not have noticed the Google BERT update. “Google introduced the BERT update to its Search ranking system last week. The addition of this new algorithm, designed to better understand what’s important in natural language queries, is a significant change. Google said it impacts 1 in 10 queries. Yet, many SEOs and many of the tracking tools did not notice massive changes in the Google search results while this algorithm rolled out in Search over the last week. The question is, Why?” Mr. Schwartz with a great explainer on the BERT update.
CNET: Google search engine will better understand natural speech, not just keywords. “Google’s search engine will now better understand your confusing search queries, the company said Friday. Google said it’s updating the tool to improve analysis of natural language. The idea is to let people type in queries that reflect how they speak in real life, instead of entering a string of keywords they think the software is more likely to understand.” I’m a little nonplussed by this; natural language searching has been a thing for a long time. Remember Ask Jeeves? Remember Electric Monk?
VentureBeat: ProBeat: Wolfram’s natural language understanding looks incredibly useful. “Wolfram Research yesterday launched Wolfram Alpha Notebook Edition for Windows, Mac, and Linux. The news largely flew under the radar, which is frankly a shame. The new tool combines Wolfram Alpha and Mathematica to give students (and teachers) a new way to build through whole computations. But it’s the natural language understanding (NLU) examples that really caught my eye.”
New York Times: A Breakthrough for A.I. Technology: Passing an 8th-Grade Science Test. “The world’s top research labs are rapidly improving a machine’s ability to understand and respond to natural language. Machines are getting better at analyzing documents, finding information, answering questions and even generating language of their own.”
EurekAlert: One class in all languages. “Now anyone from around the world can listen live to a Nobel Prize Laureate lecture or earn credits from the most reputable universities with nothing more than internet access. However, the possible information to be gained from watching and listening online is lost if the audience cannot understand the language of the lecturer. To solve this problem, scientists at the Nara Institute of Science and Technology (NAIST), Japan, presented a solution with new machine learning at the 240th meeting of the Special Interest Group of Natural Language Processing, Information Processing Society of Japan (IPSJ SIG-NL).”
Ars Technica: Microsoft open sources algorithm that gives Bing some of its smarts. “Microsoft has released today the SPTAG [Space Partition Tree and Graph] algorithm as MIT-licensed open source on GitHub. This code is proven and production-grade, used to answer questions in Bing. Developers can use this algorithm to search their own sets of vectors and do so quickly: a single machine can handle 250 million vectors and answer 1,000 queries per second. There are some samples and explanations in Microsoft’s AI Lab, and Azure will have a service using the same algorithms.”
Quartz: The emails that brought down Enron still shape our daily lives. “The Enron Corpus, as the collection is known, has been used in more than 100 projects since that research team presented it to the public in 2004. As the biggest public collection of natural written language in an organizational setting, it has been used to study everything from statistics to artificial intelligence to email attachment habits. An online art project by two Brooklyn artists will send every single one of the emails to your personal inbox, a process which (depending on the frequency of emails you request) will take anywhere from seven days to seven years.”
Towards Data Science: I trained fake news detection AI with >95% accuracy, and almost went crazy. “With so many advances in Natural Language Processing and machine learning, I thought maybe, just maybe, I could make a model that could flag news content as fake, and perhaps take a bite out of the devastating consequences of the proliferation of fake news.”
Researchers at Yahoo have developed an abuse-detecting algorithms. “The Yahoo team used a number of conventional techniques, including looking for abusive keywords, punctuation that often seemed to accompany abusive messages, and syntactic clues as to the meaning of a sentence. But the researchers also applied a more advanced approach to automated language understanding, using a way of representing the meaning of words as vectors with many dimensions.” The technique has a success rate of about 90%, which is wow.
Geektime has a writeup on a tool that translates natural language questions into SQL queries. “Kueri’s system enables developers to implant a unique search box within apps. The search box knows how to take questions from end users in natural language … and translate them into SQL queries in real time. The app can run the queries through the database and display the results to the user. In addition, in order to make it even easier for the end user, it facilitates automatic completion during typing, with completions of words and smart suggestions according to the context of the search and database.”
Google has launched a new natural languages API. “Google today announced the public beta launch of its Cloud Natural Language API, a new service that gives developers access to Google-powered sentiment analysis, entity recognition, and syntax analysis.”