Monthly Archives: August 2018

Phil 8.31.18

7:00 – 5:00 ASRC MKT

  • The lightning round slides are in!
  • Get Speaker – done
  • Get posters – done
  • Haircut – done
  • drop off DME/KLR – done
  • Under Pressure response – done, I think?
  • upload ML excel files (done) to play around with graph laplacians some more – done
  • Print out two travel packets – done
  • create shared itinerary document – started. Aaron needs to finish his part
  • From KQED Silicon Valley Conversations The Future of Music: Computer or Composer
    • Ge Wang is an Associate Professor at Stanford University in the Center for Computer Research in Music and Acoustics (CCRMA). He specializes in the art of computer music design — researching programming languages and interactive software design for music, interaction design, expressive mobile music, new performance ensembles (laptop orchestra and mobile phone orchestra), human-computer interaction, visualization (sndpeek), music game design, aesthetics of technology-mediated design, and methodologies for education at the intersection of art, engineering, and design.
    • Doug Eck is a research scientist working on Magenta, a research project exploring the role of machine learning in the process of creating art and music. Primarily this involves developing new deep learning and reinforcement learning algorithms for generating songs, images, drawings, and other materials. But it’s also an exploration in building smart tools and interfaces that allow artists and musicians to extend (not replace!) their processes using these models. Started by me in 2016, Magenta now involves several researchers and engineers from the Google Brain team as well as many others collaborating via open source. Aside from Magenta, I’m working on sequence learning models for summarization and text generation as well new ways to improve AI-generated content based on user feedback.
    • Amy X Newburg has been developing her own brand of irreverently genre-crossing works for voice, live electronics and chamber ensembles for over 25 years, known for her innovative use of live looping technology with electronic percussion, her 4-octave vocal range and her colorful — often humorous — lyrics. One of the earliest performers to work with live digital looping, Amy has presented her solo “avant-cabaret” songs at such diverse venues as the Other Minds and Bang on a Can new music festivals, the Berlin International Poetry Festival, the Wellington and Christchurch Jazz Festivals (New Zealand), the Warsaw Philharmonic Hall, electronic music festivals, colleges, rock clubs and concert halls throughout the U.S. and abroad.
  • Teens, Social Media & Technology 2018
    • YouTube, Instagram and Snapchat are the most popular online platforms among teens. Fully 95% of teens have access to a smartphone, and 45% say they are online ‘almost constantly’
  • Aaron found this: Density-functional fluctuation theory of crowds
    • A primary goal of collective population behavior studies is to determine the rules governing crowd distributions in order to predict future behaviors in new environments. Current top-down modeling approaches describe, instead of predict, specific emergent behaviors, whereas bottom-up approaches must postulate, instead of directly determine, rules for individual behaviors. Here, we employ classical density functional theory (DFT) to quantify, directly from observations of local crowd density, the rules that predict mass behaviors under new circumstances. To demonstrate our theory-based, data-driven approach, we use a model crowd consisting of walking fruit flies and extract two functions that separately describe spatial and social preferences. The resulting theory accurately predicts experimental fly distributions in new environments and provides quantification of the crowd “mood”. Should this approach generalize beyond milling crowds, it may find powerful applications in fields ranging from spatial ecology and active matter to demography and economics.
    • Here’s an interesting part: The DFFT analysis that we present is particularly powerful because it separates the influence of the environment on agents from interactions among those agents. 
      • This implies that it should (could? might?) be possible to calculate a social/environmental ratio for individual agents. High environmental are nomadic. High social are stampede-prone. Need to dig in further.
  • Mechanical Vibrations and Waves » Lecture 4: Coupled Oscillators, Normal Modes
    Lecture 4: Coupled Oscillators, Normal Modes (MIT opencourseware)

    • Prof. Lee analyzes a highly symmetric system which contains multiple objects. By physics intuition, one could identify a special kind of motion – the normal modes. He shows that there is a general strategy for solving the normal modes.
      • Every part of the system is oscillating at the same frequency and the same phase
      • Stopped at 42:07 to take a break. I think this is the right track though. Download this for the plane?
  • Chapter on normal modes

Phil 8.30.18

7:00 – 5:00  ASRC MKT

  • Target Blue Sky paper for iSchool/iConference 2019: The chairs are particularly looking for “Blue Sky Ideas” that are open-ended, possibly even “outrageous” or “wacky,” and present new problems, new application domains, or new methodologies that are likely to stimulate significant new research. 
  • I’m thinking that a paper that works through the ramifications of this diagram as it relates to people and machines. With humans that are slow responding with spongy, switched networks the flocking area is large. With a monolithic densely connected system it’s going to be a straight line from nomadic to stampede. Nomad-Flocking-Stampede2
    • Length: Up to 4 pages (excluding references)
    • Submission deadline: October 1, 2018
    • Notification date: mid-November, 2018
    • Final versions due: December 14, 2018
    • First versions will be submitted using .pdf. Final versions must be submitted in .doc, .docx or La Tex.
  • More good stuff on BBC Business Daily Trolling for Cash
    • Anger and animosity is prevalent online, with some people even seeking it out. It’s present on social media of course as well as many online forums. But now outrage has spread to mainstream media outlets and even the advertising industry. So why is it so lucrative? Bonny Brooks, a writer and researcher at Newcastle University explains who is making money from outrage. Neuroscientist Dr Dean Burnett describes what happens to our brains when we see a comment designed to provoke us. And Curtis Silver, a tech writer for KnowTechie and ForbesTech, gives his thoughts on what we need to do to defend ourselves from this onslaught of outrage.
  • Exposure to Opposing Views can Increase Political Polarization: Evidence from a Large-Scale Field Experiment on Social Media
    • Christopher Bail (Scholar)
    • There is mounting concern that social media sites contribute to political polarization by creating “echo chambers” that insulate people from opposing views about current events. We surveyed a large sample of Democrats and Republicans who visit Twitter at least three times each week about a range of social policy issues. One week later, we randomly assigned respondents to a treatment condition in which they were offered financial incentives to follow a Twitter bot for one month that exposed them to messages produced by elected officials, organizations, and other opinion leaders with opposing political ideologies. Respondents were re-surveyed at the end of the month to measure the effect of this treatment, and at regular intervals throughout the study period to monitor treatment compliance. We find that Republicans who followed a liberal Twitter bot became substantially more conservative post-treatment, and Democrats who followed a conservative Twitter bot became slightly more liberal post-treatment. These findings have important implications for the interdisciplinary literature on political polarization as well as the emerging field of computational social science.
  • Setup gcloud tools on laptop – done
  • Setup Tensorflow on laptop. Gave up un using CUDA 9.1, but got tf doing ‘hello, tensorflow’
  • Marcom meeting – 2:00
  • Get the concept of behaviors being a more scalable, dependable way of vetting information.
    • Eg Watching the DISI of outrage as manifested in trolling
      • “Uh. . . . not to be nitpicky,,,,,but…the past tense of drag is dragged, not drug.”: An overview of trolling strategies
        • Dr Claire Hardaker (Scholar) (Blog)
          • I primarily research aggression, deception, and manipulation in computer-mediated communication (CMC), including phenomena such as flaming, trolling, cyberbullying, and online grooming. I tend to take a forensic linguistic approach, based on a corpus linguistic methodology, but due to the multidisciplinary nature of my research, I also inevitably branch out into areas such as psychology, law, and computer science.
        • This paper investigates the phenomenon known as trolling — the behaviour of being deliberately antagonistic or offensive via computer-mediated communication (CMC), typically for amusement’s sake. Having previously started to answer the question, what is trolling? (Hardaker 2010), this paper seeks to answer the next question, how is trolling carried out? To do this, I use software to extract 3,727 examples of user discussions and accusations of trolling from an eighty-six million word Usenet corpus. Initial findings suggest that trolling is perceived to broadly fall across a cline with covert strategies and overt strategies at each pole. I create a working taxonomy of perceived strategies that occur at different points along this cline, and conclude by refining my trolling definition.
        • Citing papers
  • FireAnt (Filter, Identify, Report, and Export Analysis Toolkit) is a freeware social media and data analysis toolkit with built-in visualization tools including time-series, geo-position (map), and network (graph) plotting.
  • Fix marquee – done
  • Export to ppt – done!
    • include videos – done
    • Center title in ppt:
      • model considerations – done
      • diversity injection – done
  • Got the laptop running Python and Tensorflow. Had a stupid problem where I accidentally made a virtual environment and keras wouldn’t work. Removed, re-connected and restarted IntelliJ and everything is working!

Phil 8.29.18

7:00 – 4:30 ASRC MKT

  • This Is How Russian Propaganda Actually Works In The 21st Century (plus Kate Starbird’s twitter thoughts)
    • The Russian government discreetly funded a group of seemingly independent news websites in Eastern Europe to pump out stories dictated to them by the Kremlin, BuzzFeed News and its reporting partners can reveal.
  • How Right Wing is Right Wing Populism? Using multilingual CNNs on party manifestos.
    • Right wing populist parties in Europe are clearly different from other right wing parties in their rhetoric and electoral appeal. Some observers see substantive differences between right wing populists and other right wing parties, with populists supporting the welfare state and gender equality more than other right wing parties, often as part of an anti-immigration and anti-Muslim agenda. We test this claim using novel data produced by a multilingual convolutional neural net on political party platforms for the years 1990 to 2015 from the Manifesto Corpus. We find no systematic differences between right wing populists and non-populists on support for welfare and gender equality, though there is some evidence that more successful populists are more centrist.
  • Need to write up a 4 page blue sky paper for the 2019 iConference in DC
  • Realized that the poster had two herding DTW charts on the poster. Fixed and sent back. Hopefully it will get reprinted in time…
  • Uploaded the edited version and added them to the online presentation. Also saved out the mp4 files to use in the ppt version
  • Back to working on speech recognition. I’ve done a bunch of things that I’m documenting before I see if anything helped.
  • TL;DR – after much flailing, I found a page that actually helped. It’s a how-to (rather than quickstart) guide that includes a variety of interfaces including gcloud, Java and Python. And the gcloud command worked like a charm! All the flailing below is just for documentation on what NOT to do. Here’s what worked:
    PS D:\Development\Sandboxes\MapsFromPodcasts> gcloud ml speech recognize D:\Development\Sandboxes\MapsFromPodcasts\brook
    lyn.flac --language-code='en-US'
      "results": [
          "alternatives": [
              "confidence": 0.98360395,
              "transcript": "how old is the Brooklyn Bridge"

    Note that the audio file is the same as the one in the examples and is available from Google here:

  • For historical documentation of my flailing
    • First I opened a new Powershell window and re-ran the commands. Yup: Capture
    • Then I stumbled on the SDK support page and found this link to what may be the answer to the question on stackoverflow. CaptureIt says to run
      gcloud auth application-default login --scopes=,
    • Which I did, which caused a lot of things to happen Capture
    • First, I’m really wondering about this: To generate an access token for other uses, run: gcloud auth application-default print-access-token. This is used in both commands, si I’m wondering what it’s actually doing. What’s happening to this generated  token? is it being stored on my machine?
    • Second, it looks like I need to point at the [C:\Users\philip.feldman\AppData\Roaming\gcloud\application_default_credentials.json] file rather than the one in the project. That or copy to the dev location. I’m trying the former Capture
    • Then, I got this again ( Capture
    • Lastly, I upgraded because it said I could. Nothing works yet, so why not? Capture
    • That brought up a window with all this info:
      Your current Cloud SDK version is: 213.0.0
      You will be upgraded to version: 214.0.0
      │        These components will be updated.        │
      │           Name           │  Version   │   Size  │
      │ Cloud SDK Core Libraries │ 2018.08.24 │ 8.3 MiB │
      │ gcloud cli dependencies  │ 2018.08.24 │ 2.4 MiB │
      │                 These components will be installed.                 │
      │            Name            │       Version       │       Size       │
      │ Bundled Python             │                     │                  │
      The following release notes are new in this upgrade.
      Please read carefully for information about new features, breaking changes,
      and bugs fixed.  The latest full release notes can be viewed at:
      214.0.0 (2018-08-28)
        Breaking Changes
            ■ **(Cloud Bigtable)** Modified the arguments accepted by cbt
              createappprofile and cbt updateappprofile in the following ways:
              ≡ Removed etag argument from createappprofile.
              ≡ Renamed allow-transactional-writes option as transactional-writes.
              ≡ Added a force option to ignore warnings.
            ■ **(Cloud Bigtable)** Modified the specification for routing policies.
              A routing policy can be either "route-any" (previously of
              "multi_cluster_routing_use_any") or "route-to=".
            ■ **(Compute Engine)** Deprecated gcloud compute interconnects
              attachments create. Please use gcloud compute interconnects attachments
              dedicated create instead.
            ■ **(Compute Engine)** Removed deprecated --mode flag from gcloud
              compute networks create. Use --subnet-mode instead.
            ■ **(Compute Engine)** Removed deprecated gcloud compute networks
              switch-mode command. Use gcloud compute networks update
              --switch-to-custom-mode instead.
            ■ **(Compute Engine)** Removed deprecated gcloud compute xpn command
              group. Use gcloud compute shared-vpc instead.
        Cloud Bigtable
            ■ Restored the output of the cbt count command that was inadvertently
              removed in the previous release.
        Cloud Datalab
            ■ Updated the datalab component to the 20180820 release. Released
              changes are documented in its tracking issue at
        Cloud Dataproc
            ■ Added SCHEDULED_DELETE column to gcloud beta dataproc clusters list
              command output.
        Cloud Datastore Emulator
            ■ Released Cloud Datastore Emulator version 2.0.2.
              ≡ Improved backward compatibility with App Engine local development
                by keeping auto generated indexes in index file generated from
                previous runs.
        Cloud Functions
            ■ Promoted --runtime flag of gcloud functions deploy to GA.
        Compute Engine
            ■ Promoted the following flags to GA:
              ≡ --network-tier of gcloud compute <addresses|forwarding-rules>
              ≡ --default-network-tier of gcloud compute project-info update
              ≡ --network-tier of gcloud compute instances
              ≡ --network-tier of gcloud compute instance-templates create
            ■ Promoted gcloud compute instances simulate-maintenance-event to GA.
            ■ Promoted <get|set>-iam-policy and <add|remove>-iam-policy-bindings to
              beta in the following commands groups:
              ≡ gcloud compute sole-tenancy node-groups
              ≡ gcloud compute sole-tenancy node-templates
        Kubernetes Engine
            ■ Promoted --disk-type flag of gcloud container <clusters|node-pools>
              create to GA.
            ■ Promoted --default-max-pods-per-node flag of gcloud container
              clusters create to beta.
            ■ Promoted --max-pods-per-node flag of gcloud container node-pools
              create to beta.
            ■ Modified --monitoring-service flag of gcloud containers clusters
              update to enable Google Cloud Monitoring service with Kubernetes-native
              resource model.
            ■ Modified --logging-service flag of gcloud containers clusters update
              to enable Google Cloud Logging service with Kubernetes-native resource
            ■ Modified output of gcloud beta container clusters list for DEGRADED
              clusters to include reason for degradation.
            ■ Added --enable-private-nodes and --enable-private-endpoint to gcloud
              beta container clusters create.
            ■ Deprecated --private-cluster flag of gcloud beta container clusters
              create; use --enable-private-nodes instead.
          Subscribe to these release notes at
      Do you want to continue (Y/n)?  Y
      ╠═ Creating update staging area                             ═╣
      ╠═ Uninstalling: Cloud SDK Core Libraries                   ═╣
      ╠═ Uninstalling: gcloud cli dependencies                    ═╣
      ╠═ Installing: Bundled Python                               ═╣
      ╠═ Installing: Cloud SDK Core Libraries                     ═╣
      ╠═ Installing: gcloud cli dependencies                      ═╣
      ╠═ Creating backup and activating new installation          ═╣
      Performing post processing steps...done.
      Update done!
      To revert your SDK to the previously installed version, you may run:
        $ gcloud components update --version 213.0.0
      Press any key to continue . . .
    • So now lets see what happens with a restarted PowerShell
    • Nope, same problem. I also tried deleting the environment variable completely and the behavior is the same. So I don’t think that the file with the data is being sent? Capture
    • Interesting, the app-roaming file is not the same as the file that google had me generate for the text recognition getting started page: Capture

Phil 8.29.18

7:00 – 4:30 ASRC MKT

  • Editing videos
  • Need to think about short CHI paper about designing for culture/robot interactions. The trolly problem at scale? How would the sim be set up? The amount of randomness at the initial condition? Stiffness vs. connectivity? Beleif space is still important and is actually used as a concept in path planning
  • Visual Exploration and Comparison of Word Embeddings
    • Word embeddings are distributed representations for natural language words, and have been wildly used in many natural language processing tasks. The word embedding space contains local clusters with semantically similar words and meaningful directions, such as the analogy. However, there are different training algorithms and text corpora, which both have a different impact on the generated word embeddings. In this paper, we propose a visual analytics system to visually explore and compare word embeddings trained by different algorithms and corpora. The word embedding spaces are compared from three aspects, i.e., local clusters, semantic directions and diachronic changes, to understand the similarity and differences between word embeddings.
  • Much work on slides
  • Can’t get Google to recognise my account?
    curl.exe -H "Content-Type: application/json" -H "Authorization: Bearer "$(gcloud auth application-default print-access-token) -d @sync-request.json
    curl: (6) Could not resolve host: ya29.c.EloHBu32-0nBAqimi1Zumlot6rjGtGpUk27qTTESRLW4vtd1LY4ihxBIesU3ga-kmwCaM7YZS-JRo_KNjaC_bj13dWazBcKr4YtAEQYFzSpSBx3DwdS46DTt0bg
      "error": {
        "code": 403,
        "message": "The request is missing a valid API key.",
        "status": "PERMISSION_DENIED"

    No idea what host: ya29.c.EloHBu32-0nBAqimi1Zumlot6rjGtGpUk27qTTESRLW4vtd1LY4ihxBIesU3ga-kmwCaM7YZS-JRo_KNjaC_bj13dWazBcKr4YtAEQYFzSpSBx3DwdS46DTt0bg is

  • Found a problem with the poster. There are two herding DTW charts. Must be reprinted

Phil 8.27.18

7:00 – 5:00 ASRC MKT

  • Good chat with Barbara yesterday. She suggests horse racing podcasts, since the question is always the same “who’s going to win today” and the information to discuss is much more constrained. Additionally, there is the wagering information that could be used to determine the level of consensus?
  • Found an idiom translator! “Swing of the pendulum” occurs at least in French, German and Italian
  • Downloaded the new videos Need to put them in the ppt when the slides stabilize
  • Pinged Wayne about getting together today
  • Changed the questions page to have English, Italian, French and German terms for belief space
  • Another example of diversity injection (twitter)
  • Working on podcast text handling
      • Created the MapsFromPodcasts project in Development
      • Created an new key and downloaded the key json file
      • Installed Google Cloud Tools (213.0.0), following the directions of this page. Wow. Lots of stuff!
        Output folder: D:\Programs\GoogleCloudAPI
        Downloading Google Cloud SDK core.
        Extracting Google Cloud SDK core.
        Create Google Cloud SDK bat file: D:\Programs\GoogleCloudAPI\cloud_env.bat
        Installing components.
        Welcome to the Google Cloud SDK!
        Your current Cloud SDK version is: 213.0.0
        Installing components from version: 213.0.0
        | These components will be installed. |
        | Name | Version | Size |
        | BigQuery Command Line Tool | 2.0.34 | < 1 MiB |
        | BigQuery Command Line Tool (Platform Specific) | 2.0.34 | < 1 MiB |
        | Cloud SDK Core Libraries (Platform Specific) | 2018.06.18 | < 1 MiB |
        | Cloud Storage Command Line Tool | 4.33 | 3.6 MiB |
        | Cloud Storage Command Line Tool (Platform Specific) | 4.32 | < 1 MiB |
        | Cloud Tools for PowerShell | | |
        | Cloud Tools for PowerShell | | 17.9 MiB |
        | Default set of gcloud commands | | |
        | Windows command line ssh tools | | |
        | Windows command line ssh tools | 2017.09.15 | 1.8 MiB |
        | gcloud cli dependencies | 2018.08.03 | 1.3 MiB |
        For the latest full release notes, please visit:
        #= Creating update staging area =#
        #= Installing: BigQuery Command Line Tool =#
        #= Installing: BigQuery Command Line Tool (Platform Spec... =#
        #= Installing: Cloud SDK Core Libraries (Platform Specific) =#
        #= Installing: Cloud Storage Command Line Tool =#
        #= Installing: Cloud Storage Command Line Tool (Platform... =#
        #= Installing: Cloud Tools for PowerShell =#
        #= Installing: Cloud Tools for PowerShell =#
        #= Installing: Default set of gcloud commands =#
        #= Installing: Windows command line ssh tools =#
        #= Installing: Windows command line ssh tools =#
        #= Installing: gcloud cli dependencies =#
        #= Creating backup and activating new installation =#
        Performing post processing steps...
        Update done!
        This will install all the core command line tools necessary for working with
        the Google Cloud Platform.
        For more information on how to get started, please visit:
        Google Cloud SDK has been installed!



    • Google is sooooooooooooooooooooo Unix/Linux
  • Meeting with Wayne
    • Fix slides some more
    • Email about demo and poster – done

Phil 8.26.18

Listening to On Being with guest Mahzarin Banaji (Scholar)

  • The other thing that I do is to actually create inputs into my mind of my own making. I do think that in some ways our brains are simple and that they will believe that things are real even if they’re not. So, that’s what movies do. That’s what novels do for us. So what if I have a series of 1,000 pictures that rotate through on my screen saver of people who come from many parts of the world that I will never, ever see or even think about. Look, just take an example close by. I have no idea what life for a farmer in Iowa is. I bet it’s hard. I bet I have no idea what they have to deal with. I don’t think I will ever truly understand.But, right now, they are a distant group in my mind. I live in Cambridge, Massachusetts, and I don’t think about farming and farmers. If my screensaver literally just points out the existence of such people and what their issues might be, I believe that my brain is going to begin to care at some level. And if I show myself possibilities that don’t exist easily, that’s even better.
  • A nice example of diversity injection

Phil 8.24.18

7:00 – 4:00 ASRC MKT

  • Make more obvious the Inadvertent Social Information and Digital ISI
    • ISI
      • Trails
      • Visual clustering
      • Behavior around the commons (waterholes)
      • Presence of young
      • Mating behavior
      • etc.
    • DISI
      • Words and their overall source (Social media, website content, contributor content, auto-generated, etc)
      • Votes (likes, kudos, karma points)
      • Money (site income, blockchain ledger)
      • Linking (href, retweet, share)
      • Images & videos
  • Work more on behavior patterns of humans and animals
    • Highly organized (soccer match singing, marching, mass dancing events)
    • Wildebeest feeding, defending,migrating and stampeding
  • AutoKeras is a GitHub project that uses the ENAS algorithm. It can be installed using pip. Since it’s written in Keras it’s quite easy to control and play with, so you can even dive into the ENAS algorithm and try making some modifications. If you prefer TensorFlow or Pytorch, there’s also public code projects for those here and here!
  • From Zeynep’s twitter
    • So, Russian trolls amplified divisive content and helped spread vaccine misinformation.  Look, the challenge before us is to redefine *critical thinking* to include figuring out what to believe, not just how to be skeptical. Personal and institutional.
    • Weaponized Health Communication: Twitter Bots and Russian Trolls Amplify the Vaccine Debate
      •  Whereas bots that spread malware and unsolicited content disseminated antivaccine messages, Russian trolls promoted discord. Accounts masquerading as legitimate users create false equivalency, eroding public consensus on vaccination.
  • Trying to decode podcasts. Here’s my test , and here are the results from Google speech-to-text:
    • We were talking about the choices of who’s you can keep two of these three, I guess Adonis Alexander is along for the ride, huh? I thought I was about to I didn’t know I haven’t I haven’t sent it to him. Well, has he been out there? They might missing some guys got to hand. I kept thinking like if to say, they weren’t having these injuries. Like if they have like us to say, okay, but they have these reason iron Marshall and maybe he maybe he’s not available week one, but they don’t want to put them on IR prn’s things up. So maybe they have to add another running back like you so you have to create a roster spot I could imagine this is just speculation Alexander. Somehow gets the mysterious injury to put them on I are clearly my keys ready, right and they they would have five cornerbacks otherwise and you know, yeah, if you’re not going to be ready to go, but you may have to you know, go get okay. Yeah. I mean the he’s he’s a guy that I think is on based on these the way the wrong.
    • It’s pretty good as long as people aren’t stepping over each other verbally.
    • Good enough to try, I guess. Noisy data is life, right? Look for the bigger signal.
  • Here’s my current plan. It’s a half-assed first approach, but it should provide some insight.
    1. Download a season of a sports podcast and put each podcast into it’s own document Here’s the tutorial for Speech-to-text with REST
    2. Use Corpus Manager to convert, using BOW and create an ignore list for common words like “the”
    3. then read all the docs into LMN
    4. Then set the weight of each successive document (in time) so that its top
    5. Take the top ten words and save them to a file
    6. Try building a map

Phil 8.23.18

7:00 – 5:30 ASRC MKT


  • Slides
    • Groups/tribes stay the same, but the topics change
    • Past polarizing topics:
      • Confederate statues
      • Kneeling for the national anthem
      • #blacklivesmatter
      • Hoodies
      • Crack cocaine
      • 1968 Olympics Black Power salute
      • Alabama bus boycott
    • Stiffening a group creates a stampede (In-group high SIH)
    • Adding group-invisible diversity disrupts the velocity and direction of a stampede
    • Arendt/Moscovici slide “So we’re doomed, right! Except…”
    • See what velocity of the disrupted stampede looks like
  • Why Trump Supporters Believe He Is Not Corrupt
    • The answer may lie in how Trump and his supporters define corruption. In a forthcoming book titled How Fascism Works, the Yale philosophy professor Jason Stanley makes an intriguing claim. “Corruption, to the fascist politician,” he suggests, “is really about the corruption of purity rather than of the law. Officially, the fascist politician’s denunciations of corruption sound like a denunciation of political corruption. But such talk is intended to evoke corruption in the sense of the usurpation of the traditional order.”
  • Climate science proposals are being reviewed by Ryan Zinke’s old football buddy. Seriously.
    • But what if the corruption isn’t hidden at all, but right out in the open? What if, when it’s identified, the perpetrator doesn’t apologize, or demonstrate any remorse or shame, and there’s no punishment? What then? We don’t really have good narratives around what happens in that situation, which is why the Trump administration so often leaves us sputtering and gawking. It can’t just be a motley collection of incompetent grifters, each misruling their own little fiefdom, trying to stay in their boss’s good graces, succeeding less through wits than a congenital lack of shame and the unstinting institutional support of GOP donors. Can it?

Phil 8.22.18

7:00 – 4:00 ASRC MKT

Phil 8.21.18

7:00 – 3:00 ASRC MKT

  • Rework the slides
    • Explicit introduction, lit review, methods, results, conclusion and discussion slides
    • Slide for the difference between opinion dynamics & consensus formation as a static end  and part of a dynamic process. (Tribe membership may be static, belief of the tribe is highly dynamic. It’s the story for the group)
    • Revisit stampede/flock/nomad slide in the conclusions
    • Lose the following slides:
      • Belief space
      • Theory slide replace with a slide that breaks out the to knobs of dimension reduction and social influence horizons. The slide is called “the simple trick” Explain how herding affects these knobs by presenting simple issues and making the network stiffer through weight and connection
    • Get rid of optical polarization
  • Fanning the Flames of Hate: Social Media and Hate Crime
    • This paper investigates the link between social media and hate crime using Facebook data. We study the case of Germany, where the recently emerged right-wing party Alternative für Deutschland (AfD) has developed a major social media presence. We show that right-wing anti-refugee sentiment on Facebook predicts violent crimes against refugees in otherwise similar municipalities with higher social media usage. To further establish causality, we exploit exogenous variation in major internet and Facebook outages, which fully undo the correlation between social media and hate crime. We further find that the effect decreases with distracting news events; increases with user network interactions; and does not hold for posts unrelated to refugees. Our results suggest that social media can act as a propagation mechanism between online hate speech and real-life violent crime.
  • Facebook is rating the trustworthiness of its users on a scale from zero to 1
    • Facebook has begun to assign its users a reputation score, predicting their trustworthiness on a scale from zero to 1.
    • Tessa Lyons, product manager who is in charge of fighting misinformation (video)
  • Social Science One
    • implements a new type of partnership between academic researchers and private industry to advance the goals of social science in understanding and solving society’s greatest challenges. The partnership enables academics to analyze the increasingly rich troves of information amassed by private industry in responsible and socially beneficial ways. It ensures the public maintains privacy while gaining societal value from scholarly research. And it enables firms to enlist the scientific community to help them produce social good, while protecting their competitive positions.
  • Lost Causes Is this fashion in economic theory (found via Twitter)?Causal
  • Poster printing – UMBC Commonvision

Phil 8.19.18

7:00 – 5:30 ASRC MKT

  • Had a thought that the incomprehension that comes from misalignment that Stephens shows resembles polarizing light. I need to add a slider that enables influence as a function of alignment. Done
    • Getting the direction cosine between the source and target belief
      double interAgentDotProduct = unitOrientVector.dotProduct(otherUnitOrientVector);
      double cosTheta = Math.min(1.0, interAgentDotProduct);
      double beliefAlignment = Math.toDegrees(Math.acos(cosTheta));
      double interAgentAlignment = (1.0 - beliefAlignment/180.0);
    • Adding a global variable that sets how much influence (0% – 100%) influence from an opposing agent. Just setting it to on/off, because the effects are actually pretty subtle
  • Add David’s contributions to slide one writeup – done
  • Start slide 2 writeup
  • Find casters for Dad’s walker
  • Submit forms for DME repair
    • Drat – I need the ECU number
  • Practice talk!
    • Need to reduce complexity and add clearly labeled sections, in particular methods
  • I need to start paying attention to attention
  • Also, keeping this on the list How social media took us from Tahrir Square to Donald Trump by Zeynep Tufekci
  • Social Identity Threat Motivates Science – Discrediting Online Comments
    • Experiencing social identity threat from scientific findings can lead people to cognitively devalue the respective findings. Three studies examined whether potentially threatening scientific findings motivate group members to take action against the respective findings by publicly discrediting them on the Web. Results show that strongly (vs. weakly) identified group members (i.e., people who identified as “gamers”) were particularly likely to discredit social identity threatening findings publicly (i.e., studies that found an effect of playing violent video games on aggression). A content analytical evaluation of online comments revealed that social identification specifically predicted critiques of the methodology employed in potentially threatening, but not in non-threatening research (Study 2). Furthermore, when participants were collectively (vs. self-) affirmed, identification did no longer predict discrediting posting behavior (Study 3). These findings contribute to the understanding of the formation of online collective action and add to the burgeoning literature on the question why certain scientific findings sometimes face a broad public opposition.

Phil 8.18.18

This looks good:

  • Created almost 25 years ago, when the web was in its infancy, Propaganda Critic is dedicated to promoting techniques of propaganda analysis among critically minded citizens.

    In 2018, realizing that traditional approaches to propaganda analysis were not well-suited for making sense out of our contemporary political crisis, we completely overhauled Propaganda Critic to take into account the rise of ‘computational propaganda.’ In addition to updating all of the original content, we added nearly two dozen new articles exploring the rise of computational propaganda, explaining recent research on cognitive biases that influence how we interpret and retain information, and presenting recent case studies of how propaganda techniques have been used to disrupt democracy around the world.

Continuing to work on the SASO writeup – it’s coming along. Slower than I’d like…

This is just too good:

  • Data Organization in Spreadsheets
    • Spreadsheets are widely used software tools for data entry, storage, analysis, and visualization. Focusing on the data entry and storage aspects, this article offers practical recommendations for organizing spreadsheet data to reduce errors and ease later analyses. The basic principles are: be consistent, write dates like YYYY-MM-DD, do not leave any cells empty, put just one thing in a cell, organize the data as a single rectangle (with subjects as rows and variables as columns, and with a single header row), create a data dictionary, do not include calculations in the raw data files, do not use font color or highlighting as data, choose good names for things, make backups, use data validation to avoid data entry errors, and save the data in plain text files.

Phil 8.17.18

7:00 – 4:30 ASRC MKT

Phil 8.16.18

7:00 – 4:30 ASRC MKT

  • R2D3 is an experiment in expressing statistical thinking with interactive design. Find us at @r2d3usR2D3
  • Foundations of Temporal Text Networks
    • Davide Vega (Scholar)
    • Matteo Magnani (Scholar)
    • Three fundamental elements to understand human information networks are the individuals (actors) in the network, the information they exchange, that is often observable online as text content (emails, social media posts, etc.), and the time when these exchanges happen. An extremely large amount of research has addressed some of these aspects either in isolation or as combinations of two of them. There are also more and more works studying systems where all three elements are present, but typically using ad hoc models and algorithms that cannot be easily transferred to other contexts. To address this heterogeneity, in this article we present a simple, expressive and extensible model for temporal text networks, that we claim can be used as a common ground across different types of networks and analysis tasks, and we show how simple procedures to produce views of the model allow the direct application of analysis methods already developed in other domains, from traditional data mining to multilayer network mining.
      • Ok, I’ve been reading the paper and if I understand it correctly, it’s pretty straightforward and also clever. It relates a lot to the way that I do term document matrices, and then extends the concept to include time, agents, and implicitly anything you want to. To illustrate, here’s a picture of a tensor-as-matrix: tensorIn2DThe important thing to notice is that there are multiple dimensions represented in a square matrix. We have:
        • agents
        • documents
        • terms
        • steps
      • This picture in particular is of an undirected adjacency matrix, but I think there are ways to handle in-degree and out-degree, though I think that’s probably better handled by having one matrix for indegree and one for out.
      • Because it’s a square matrix, we can calculate the steps between any node that’s on the matrix, and the centrality, simply by squaring the matrix and keeping track of the steps until the eigenvector settles. We can also weight nodes by multiplying that node’s row and column by the scalar. That changes the centrality, but ot the connectivity. We can also drop out components (steps for example) to see how that changes the underlying network properties.
      • If we want to see how time affects the development of the network, we can start with all the step nodes set to a zero weight, then add them in sequentially. This means, for example, that clustering could be performed on the nonzero nodes.
      • Some or all of the elements could be factorized using NMF, resulting in smaller, faster matrices.
      • Network embedding could be useful too. We get distances between nodes. And this looks really important: Network Embedding as Matrix Factorization: Unifying DeepWalk, LINE, PTE, and node2vec
      • I think I can use any and all of the above methods on the network tensor I’m describing. This is very close to a mapping solution.
  • The Shifting Discourse of the European Central Bank: Exploring Structural Space in Semantic Networks (cited by the above paper)
    • Convenient access to vast and untapped collections of documents generated by organizations is a valuable resource for research. These documents (e.g., Press releases, reports, speech transcriptions, etc.) are a window into organizational strategies, communication patterns, and organizational behavior. However, the analysis of such large document corpora does not come without challenges. Two of these challenges are 1) the need for appropriate automated methods for text mining and analysis and 2) the redundant and predictable nature of the formalized discourse contained in these collections of texts. Our article proposes an approach that performs well in overcoming these particular challenges for the analysis of documents related to the recent financial crisis. Using semantic network analysis and a combination of structural measures, we provide an approach that proves valuable for a more comprehensive analysis of large and complex semantic networks of formal discourse, such as the one of the European Central Bank (ECB). We find that identifying structural roles in the semantic network using centrality measures jointly reveals important discursive shifts in the goals of the ECB which would not be discovered under traditional text analysis approaches.
  • Comparative Document Analysis for Large Text Corpora
    • This paper presents a novel research problem, Comparative Document Analysis (CDA), that is, joint discovery of commonalities and differences between two individual documents (or two sets of documents) in a large text corpus. Given any pair of documents from a (background) document collection, CDA aims to automatically identify sets of quality phrases to summarize the commonalities of both documents and highlight the distinctions of each with respect to the other informatively and concisely. Our solution uses a general graph-based framework to derive novel measures on phrase semantic commonality and pairwise distinction, where the background corpus is used for computing phrase-document semantic relevance. We use the measures to guide the selection of sets of phrases by solving two joint optimization problems. A scalable iterative algorithm is developed to integrate the maximization of phrase commonality or distinction measure with the learning of phrase-document semantic relevance. Experiments on large text corpora from two different domains—scientific papers and news—demonstrate the effectiveness and robustness of the proposed framework on comparing documents. Analysis on a 10GB+ text corpus demonstrates the scalability of our method, whose computation time grows linearly as the corpus size increases. Our case study on comparing news articles published at different dates shows the power of the proposed method on comparing sets of documents.
  • Social and semantic coevolution in knowledge networks
    • Socio-semantic networks involve agents creating and processing information: communities of scientists, software developers, wiki contributors and webloggers are, among others, examples of such knowledge networks. We aim at demonstrating that the dynamics of these communities can be adequately described as the coevolution of a social and a socio-semantic network. More precisely, we will first introduce a theoretical framework based on a social network and a socio-semantic network, i.e. an epistemic network featuring agents, concepts and links between agents and between agents and concepts. Adopting a relevant empirical protocol, we will then describe the joint dynamics of social and socio-semantic structures, at both macroscopic and microscopic scales, emphasizing the remarkable stability of these macroscopic properties in spite of a vivid local, agent-based network dynamics.
  • Tensorflow 2.0 feedback request
    • Shortly, we will hold a series of public design reviews covering the planned changes. This process will clarify the features that will be part of TensorFlow 2.0, and allow the community to propose changes and voice concerns. Please join if you would like to see announcements of reviews and updates on process. We hope to gather user feedback on the planned changes once we release a preview version later this year.

Phil 8.12.18

7:00 – 4:00 ASRC MKT

  • Having an interesting chat on recommenders with Robin Berjon on Twitter
  • Long, but looks really good Neural Processes as distributions over functions
    • Neural Processes (NPs) caught my attention as they essentially are a neural network (NN) based probabilistic model which can represent a distribution over stochastic processes. So NPs combine elements from two worlds:
      • Deep Learning – neural networks are flexible non-linear functions which are straightforward to train
      • Gaussian Processes – GPs offer a probabilistic framework for learning a distribution over a wide class of non-linear functions

      Both have their advantages and drawbacks. In the limited data regime, GPs are preferable due to their probabilistic nature and ability to capture uncertainty. This differs from (non-Bayesian) neural networks which represent a single function rather than a distribution over functions. However the latter might be preferable in the presence of large amounts of data as training NNs is computationally much more scalable than inference for GPs. Neural Processes aim to combine the best of these two worlds.

  • How The Internet Talks (Well, the mostly young and mostly male users of Reddit, anyway)
    • To get a sense of the language used on Reddit, we parsed every comment since late 2007 and built the tool above, which enables you to search for a word or phrase to see how its popularity has changed over time. We’ve updated the tool to include all comments through the end of July 2017.
  • Add breadcrumbs to slides
  • Download videos – done! Put these in the ppt backup
  • Fix the DTW emergent population chart on the poster and in the slides. Print!
  • Set up the LaTex Army BAA framework
  • Olsson
  • Slide walkthough. Good timing. Working on the poster some more AdversarialHerding2