Bayesian inference methods for probabilistic topic models can quantify uncertainty in the parameters, which has primarily been used to increase the robustness of parameter estimates. In this work, we explore other rich information that can be obtained by analyzing the posterior distributions in topic models. Experimenting with latent Dirichlet allocation on two datasets, we propose ideas incorporating information about the posterior distributions at the topic level and at the word level. At the topic level, we propose a metric called topic stability that measures the variability of the topic parameters under the posterior. We show that this metric is correlated with human judgments of topic quality as well as with the consistency of topics appearing across multiple models. At the word level, we experiment with different methods for adjusting individual word probabilities within topics based on their uncertainty. Humans prefer words ranked by our adjusted estimates nearly twice as often when compared to the traditional approach. Finally, we describe how the ideas presented in this work could potentially applied to other predictive or exploratory models in future work.
Zhang, Yuanzhe (Institute of Automation, Chinese Academy of Sciences) | He, Shizhu (Institute of Automation, Chinese Academy of Sciences) | Liu, Kang (Institute of Automation, Chinese Academy of Sciences) | Zhao, Jun (Institute of Automation, Chinese Academy of Sciences)
As the amount of knowledge bases (KBs) grows rapidly, the problem of question answering (QA) over multiple KBs has drawn more attention. The most significant distinction between multiple KB-QA and single KB-QA is that the former must consider the alignments between KBs. The pipeline strategy first constructs the alignments independently, and then uses the obtained alignments to construct queries. However, alignment construction is not a trivial task, and the introduced noises would be passed on to query construction. By contrast, we notice that alignment construction and query construction are interactive steps, and jointly considering them would be beneficial. To this end, we present a novel joint model based on integer linear programming (ILP), uniting these two procedures into a uniform framework. The experimental results demonstrate that the proposed approach outperforms state-of-the-art systems, and is able to improve the performance of both alignment construction and query construction.
Its venerable phone line wasn't the only newly minted product Apple showed off at the iPhone 8 event on Tuesday. Eddie Cue announced onstage that the company will expand availability of its TV app to seven new countries by the end of the year and will be adding local news and sports programming as well. The TV app will be available in Australia and Canada next month, the spread to Germany, France, Sweden, Norway and the UK by the end of the year. US sports fans (that is, those that live in the country), will be able to track their favorite teams and have Apple TV push an on-screen notification whenever a game starts. By the end of the year, Apple also announced that users will be able to ask Siri directly to switch to a game.
Funny artificial intelligence is nothing new. We might be used to Siri's bad jokes on our iPhones or messing with Amazon's Alexa. And recently an AI robot Sophia tried out a pickup line on Charlie Rose on 60 Minutes last week. Now Google has joined the trend of human-like artificial intelligence to make AI a little more fun. Its new assistant was developed in part by comedy writers from Pixar and The Onion, the satirical newspaper that is sometimes just too honest for these crazy times we live in, according to The Wall Street Journal.
Amazon has brought Apple Music to Alexa smart speakers for the first time in the UK. The feature was already available in the US but is slowly rolling out across the world. It means that people are no longer limited to only Amazon Music and Spotify as ways for Alexa to play music. Apple Music will work like those same services, allowing people to ask for a song or radio station and have it play from Apple Music. We'll tell you what's true.