Author Archives: pgfeldman

Phil 4.26.16

7:00 – 4:00 VTX

  • Reading through (and coding) A Fistfull of Bitcoins. In the ‘duh’ department, I realize that it should be possible to pay anonymous sources using BC since they both rely on the same mechanism. So when you submit a story, you can also use a bitcoin address. It would help in tracking users, that’s for sure. If you want to associate a bitcoin address at a later time, then a more detailed biometric analysis would have to take place. Maybe a game. Also, users should be able to create a BC address alias. These would have to be unique across the system(? Is this really true?), but that’s kind of like user name, so there are issues….
  • Worked on JavaFX layout issues need to figure out how to get the grid to scale? Or maybe use anchor points. More tomorrow.
  • Sprint retrospective
  • Presented the tool. Need to add users.

Phil 4.25.16

5:30 – 4:00 VTX

  • Saw this on Twitter about visualizing networks with D3
  • Working my way through the JavaFX tutorial. It is a lot like a blend of Flex and a rethought Swing. Nice, actually…
  • Here is the list of stock components
  • Starting with the ope file dialog – done.
  • Yep, there’s a spinner. And here’s dials and knobs
  • And here’s how to do a word cloud.
  • Here’s a TF-IDF implementation in JAVA. Need to build some code that reads in from our ‘negative match’ ‘positive match’ results and start to get some data driven terms
  • Tregex is a utility for matching patterns in trees, based on tree relationships and regular expression matches on nodes (the name is short for “tree regular expressions”). Tregex comes with Tsurgeon, a tree transformation language. Also included from version 2.0 on is a similar package which operates on dependency graphs (class SemanticGraph, calledsemgrex).
  • Semgrex
  • Sprint review
    • Google CSEs
      • Switched over from my personal CSEs to Vistronix CSEs
      • Added VCS rep for CSEs
      • Figured out how to save out and load CSE from XML
      • Added a few more CSEs ONLY_NET, MOBY_DICK
      • Wrote up care and feeding document for Confluence
      • Added blacklists
    • Rating App
      • Re-rigged the JPA classes to be Ontology-agnostic Version 2 of nearly everything)
      • Upped my JQL game to handle SELECT IN WHERE precompiled queries
      • Reading in VA and PA data now
      • Added the creation of a text JSON object that formalizes the rating of a flag
      • Got hooked up to the Talend DB!!!
      • Deployed initial version(s)
      • Added backlink logging using SemRush
    • Future work
      • Developed Excel ingest
      • Still working on PDF and Word ingest

Phil 4.22.16

7:00 – 4:30 VTX

  • Had a thought going to sleep last night that it would be interesting to see the difference between a ‘naive’ ranking based on the number of quotes vs. PageRank. Pretty much as soon as I got up, I pulled down the spreadsheet and got the lists. It’s in the previous post, but I’ll pot them here too:
    • Sorted from most to least quotes
      P61: A Survey on Assessment and Ranking Methodologies for User-Generated Content on the Web.pdf
      P13: The Egyptian Blogosphere.pdf
      P10: Sensing_And_Shaping_Emerging_Conflicts.pdf
      P85: Technology Humanness and Trust-Rethinking Trust in Technology.pdf
      P 5: Saracevic_relevance_75.pdf
      P 1: Social Media and Trust during the Gezi Protests in Turkey.pdf
      P77: The Law of Group Polarization.pdf
      P43: On the Accuracy of Media-based Conflict Event Data.pdf
      System Trust
      P37: Security-control methods for statistical databases – a comparative study.pdf
    • Sorted on Page Rank eigenvector
      P85: Technology Humanness and Trust-Rethinking Trust in Technology.pdf
      System Trust
      Social Trust
      P61: A Survey on Assessment and Ranking Methodologies for User-Generated Content on the Web.pdf
      P84: What is Trust_ A Conceptual Analysis–AMCIS-2000.pdf
      P 1: Social Media and Trust during the Gezi Protests in Turkey.pdf
      Credibility Cues
      P13: The Egyptian Blogosphere.pdf
      P10: Sensing_And_Shaping_Emerging_Conflicts.pdf
      P82: The ‘like me’ framework for recognizing and becoming an intentional agent.pdf
  • To me it’s really interesting how much better the codes are mixed in to the results. I actually thought it could be the other way, since the codes are common across many papers. Also, the concepts of System Trust, Social Trust and Credibility  Cues very much became a central point in my mind as I worked through the papers.
  • A second thought, which is the next step in the research, is to see ho weighting affects relationships. Right now, the the papers and codes are weighted by the number of quotes. What happens when all the weights are normalized (set to 1.0)?. And then there is the setup of the interactivity. With zero optimizations, this took 4.2 seconds to calculate on a modern laptop. Not sliderbar rates, but change a (some?) values and click a ‘run’ button.
  • So, moving forward, the next steps are to create the Swing App that will:
    • read in a spreadsheet (xls and xlsx)
    • Write out spreadsheets (page containing the data information
      • File
      • User
      • Date run
      • Settings used
    • allow for manipulation of row and column values (in this case, papers and codes, but the possibilities are endless)
      • Select the value to manipulate (reset should be an option)
      • Spinner/entry field to set changes (original value in label)
      • ‘Calculate’ button
      • Sorted list(s) of rows and columns. (indicate +/- change in rank)
    • Reset all button
    • Normalize all button
  • I’d like to do something with the connectivity graph. Not sure what yet.
  • And I think I’ll do this in JavaFX rather than Swing this time.
  • Huh. JavaFX Scene Builder is no longer supported by Oracle. Now it’s a Gluon project.
  • Documentation still seems to be at Oracle though
  • Spent most of the day seeing what’s going on with the Crawl. Turns out it was bad formatting on the terms?

Phil 4.21.16

7:00 – VTX

  • A little more bitcoin
  • Installed *another* new Java 1.8.0_92
  • Discovered the arXiv API page. This might be very helpful. I need to dig into it a bit.
  • Testing ranking code. I hate to say this, but if it works I think I’m going to write *another* Swing app to check interactivity rates. Which means I need to instrument the matrix calculations for timing.
  • Ok, the rank table is consistent across all columns. In my test code, the eigenvector stabilizes after 5 iterations: 
    initial
 , col1, col2, col3, col4,
row1, 11, 21, 31, 41,
row2, 12, 22, 32, 42,
row3, 13, 23, 33, 43,

derived
, row1, row2, row3, col1, col2, col3, col4,
row1, 1, 0, 0, 0.26, 0.49, 0.72, 0.95,
row2, 0, 1, 0, 0.28, 0.51, 0.74, 0.98,
row3, 0, 0, 1, 0.3, 0.53, 0.77, 1,
col1, 0.26, 0.28, 0.3, 1, 0, 0, 0,
col2, 0.49, 0.51, 0.53, 0, 1, 0, 0,
col3, 0.72, 0.74, 0.77, 0, 0, 1, 0,
col4, 0.95, 0.98, 1, 0, 0, 0, 1,

rank
, row1, row2, row3, col1, col2, col3, col4,
row1, 0.61, 0.62, 0.64, 0.22, 0.41, 0.59, 0.78,
row2, 0.62, 0.65, 0.67, 0.23, 0.42, 0.61, 0.8,
row3, 0.64, 0.67, 0.69, 0.24, 0.43, 0.63, 0.83,
col1, 0.22, 0.23, 0.24, 0.08, 0.15, 0.22, 0.29,
col2, 0.41, 0.42, 0.43, 0.15, 0.27, 0.4, 0.52,
col3, 0.59, 0.61, 0.63, 0.22, 0.4, 0.58, 0.76,
col4, 0.78, 0.8, 0.83, 0.29, 0.52, 0.76, 1,

EigenVec
row1, 1, 0.71, 0.62, 0.61, 0.61, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
row2, 0, 0.46, 0.61, 0.62, 0.62, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
row3, 0, 0.48, 0.63, 0.64, 0.64, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
col1, 0.26, 0.13, 0.21, 0.22, 0.22, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
col2, 0.49, 0.25, 0.38, 0.41, 0.41, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
col3, 0.72, 0.37, 0.55, 0.59, 0.59, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
col4, 0.95, 0.49, 0.73, 0.78, 0.78, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
  • And after a lot of banging away, here’s my lit review in PageRank: pageRank
  • And here’s the difference between PageRank and sorting based on number of quotes:
  • Page Rank
    P85: Technology Humanness and Trust-Rethinking Trust in Technology.pdf
    System Trust
    Social Trust
    P61: A Survey on Assessment and Ranking Methodologies for User-Generated Content on the Web.pdf
    P84: What is Trust_ A Conceptual Analysis–AMCIS-2000.pdf
    P 1: Social Media and Trust during the Gezi Protests in Turkey.pdf
    Credibility Cues
    P13: The Egyptian Blogosphere.pdf
    P10: Sensing_And_Shaping_Emerging_Conflicts.pdf
    P82: The ‘like me’ framework for recognizing and becoming an intentional agent.pdf
  • Sorted from most to least quotes
    P61: A Survey on Assessment and Ranking Methodologies for User-Generated Content on the Web.pdf
    P13: The Egyptian Blogosphere.pdf
    P10: Sensing_And_Shaping_Emerging_Conflicts.pdf
    P85: Technology Humanness and Trust-Rethinking Trust in Technology.pdf
    P 5: Saracevic_relevance_75.pdf
    P 1: Social Media and Trust during the Gezi Protests in Turkey.pdf
    P77: The Law of Group Polarization.pdf
    P43: On the Accuracy of Media-based Conflict Event Data.pdf
    System Trust
    P37: Security-control methods for statistical databases – a comparative study.pdf

Phil 4.20.16

7:00 – 4:00 VTX

  • Read a little more of the BitCoin article. Nice description of the blockchain.
  • Generated a fresh Excel matrix of codes and papers. I excluded the ‘meta’ codes (Definitions, Methods, etc) and all the papers that have zero quotes
  • Duke Ellington & His Orchestra make great coding music.
  • Installed new Java
  • Well drat. I can’t increase the size of an existing matrix. Will have to return a new one.
  • Yay! 
    FileUtils.getInputFileName() opening: mat1.xls
initial
 , col1, col2, col3, col4, 
row1, 11.0, 21.0, 31.0, 41.0, 
row2, 12.0, 22.0, 32.0, 42.0, 
row3, 13.0, 23.0, 33.0, 43.0, 
done calculating
done creating
derived
 , row1, row2, row3, col1, col2, col3, col4, 
row1, 0.0, 0.0, 0.0, 11.0, 21.0, 31.0, 41.0, 
row2, 0.0, 0.0, 0.0, 12.0, 22.0, 32.0, 42.0, 
row3, 0.0, 0.0, 0.0, 13.0, 23.0, 33.0, 43.0, 
col1, 11.0, 12.0, 13.0, 0.0, 0.0, 0.0, 0.0, 
col2, 21.0, 22.0, 23.0, 0.0, 0.0, 0.0, 0.0, 
col3, 31.0, 32.0, 33.0, 0.0, 0.0, 0.0, 0.0, 
col4, 41.0, 42.0, 43.0, 0.0, 0.0, 0.0, 0.0,
  • Need to set Identity and other housekeeping.
  • Added ‘normalizeByMatrix that sets the entire matrix on a unit scale
  • Need to have a calcRank function that squares and normalizes until the difference between output eigenvectors are below a certain threshold or a limit of iterations. Done?

Phil 4.19.16

7:00 – 4:00 VTX

  • No reading today! Actually, scanning the Bitcoin acm article, but the goal is to get the PageRank code working with the excel outputs from Atlas and then be able to re-weight the eigenvectors. Then we’ll see if that can be used for collapsing the codes.
  • Reading in spreadsheets!
  • Working on buildSymmetricMatrix that will create a matrix that has all the row and columns represented on each axis of the matrix. it looks(?) like setSubMatrix will allow this…
  • Got the symmetric double[][] built and validated in the debugger. tomorrow we put this all together.
  • Walked through my ML concept for document filtering with Mike G and Aaron in two unrelated conversations.
  • Checked in John Robert’s ONLY_NET CSE

Phil 4.18.16

7:00 – 11:00 VTX

  • Timesheets
  • Sent Aaron a status since I’ll be in class for scrum
  • Continuing The ‘like me’ framework for recognizing and becoming an intentional agent
    • Page 12, at the end of General Discussion: Based on the ‘like me’ perception of others’ actions, social encounters are more interpretable than supposed by classical ‘solipsism’ theories such as Piaget’s. Infants can use themselves as a framework for understanding the subjectivity of others and reciprocally learn about the possibilities inherent in their own actions by observing the actions of others. Through social interaction with other intentional agents who are viewed as ‘like me,’ infants develop a richer social cognition
      • So we’re wired to trust that other people’s(!) actions will be like our own. We learn early (only?) behavior through empathy. That sets the pattern we use going forwards. And it is also wired into us. This suggests two things. First, it’s hackable. Second, there are affordances for social and system trust that should be used in the designs of systems that one, may have a ‘social back end’ (IR/Search/GPS), and two, systems that appear to be social need to be carefully constructed so that human trusting intention isn’t short-circuited.
  • Continuing to work on Adjacency and PageRank java code. Working from home today, so that means updating, checking out and fixing what’s broke.
    • Built test matrices from codes and docs
    • IntelliJ housekeeping. New version! Restarts…
    • And the code compiles!
    • Need to read from excel
    • Need a ‘build symmetric matrix’ method

Phil 4.15.16

7:00 – 4:30 VTX

  • Good meeting with Wayne yesterday evening
  • Tensorflow playground
  • Continuing The ‘like me’ framework for recognizing and becoming an intentional agent
    • Page 4: Based on the ‘like me’ framework, I hypothesized that it would be possible to demonstrate such tool-use learning at younger ages by transforming the situation. Instead of having the infant sit across the table from the adult, I had them sit side-by-side. In that way the adult’s actions could more easily serve as a blueprint for the child’s own action plans. Recent brain imaging studies with adults show the facilitative effects of seeing a to-be-imitated action from one’s own point of view (Jackson, Meltzoff, & Decety, 2006).
    • Page 5:This study was the first to show infants how to use complex tools ‘from their own perspective.’ Sitting shoulder-to-shoulder with the child closes the gap between the perceived and executed actions. The model becomes more ‘like me.’ 
      • Eyewitness value, photos and images all come from a ‘like me’ framework. As much as possible, we are looking out of the eyes of the witness. This high level of credibility traces all the way back to infancy. Wow. On a related note, this has implications for news reporting using VR.
    • Page 6: Evidently, young toddlers can understand our goals even if we fail to fulfill them. In another study (Meltzoff, 1995; Experiment 2), it was shown that infants did not reenact the target act if  they saw a mechanical device rather than a person performing the ‘slipping’ movements. The device did not look human and had poles as arms and pincers instead of fingers, but it traced the same spatiotemporal pattern as did the person’s yanking. Infants did not pull apart the dumbbell at any higher than baseline levels in this case. They did, however, correctly perform the target act in another condition in which the mechanical device succeeded in pulling apart the dumbbell. This makes sense, because in the case of success the object transformation is visible (it is pulled apart), but in the case of the unsuccessful attempt, there is no object transformation, only a ‘slipping’ motion that has to be interpreted at a different level.
      • Does this mean that we have a ‘wired-in’ model of the intention of others?
    • Page 7: Persistence and emotions as markers of infants’ intention—In further work, I showed 18-month-olds (N = 33) the standard unsuccessful-attempt display, but handed them a trick toy. The toy had been surreptitiously glued shut before the study began. When infants picked it up and attempted to pull it apart, their hands slipped off the ends. This, of course, matched the surface behavior of the adult. The question was whether this imitation of the adults’ behavior satisfied the infants. It did not. When infants matched the surface behavior of the adult, they did not terminate their behavior. They repeatedly grabbed the toy, yanked on it in different ways, and appealed to the adult for help by looking and vocalizing. About 90% (20/23) of those who tried to pull apart the object immediately stared at the adult after they failed to do so (mean latency = 1.74 s). Why were they appealing for help? They had matched the adult’s surface behavior. Evidently, they were striving toward something else: the adult’s goals, not his literal behavior
      • Definately a model of something… And a goal.
    • Page 7: We also conducted related neuroscience work in adults. The results reveal that neural structures known to be involved in adult theory-of-mind tasks (medial prefrontal cortex) are activated in tasks requiring adults to infer unconsummated goals in basic action tasks (Chaminade, Meltzoff, & Decety, 2002; see also Reid, Csibra, Belsky, & Johnson, 2007, for related work). This suggests a tie between the processing of action sequences in terms of goals and more sophisticated aspects of social cognition.
    • Page 7: Our adult commonsense psychology includes a distinction between the types of entities that are accorded goals and intentions and those that are not. We ascribe a goal to the archer not to the arrow that reaches (or misses) the target
      • That’s a fundamental ‘humanness’ definition that Social Trust depends on. If the inferred goals are trustworthy, then slips in behavior are discounted.
    • Page 7: I am currently exploring whether mechanical devices such as social robots can be treated as ‘like me’ based on bodily structure and/ or the type of behavior they exhibit, prompting action imitation by the infant. Preliminary results suggest so.
  • —————–
  • Updated the deployable RatingApp.exe. Asked Andy to set up a Skype meeting so I can demo.
  • Presented and deployed.
  • Made a new CSE that only points to the online Moby Dick, that can be used for query testing.

Phil 4.14.16

7:00 – 3:30 VTX

  • Continuing The ‘like me’ framework for recognizing and becoming an intentional agent
  • Page 2: Perception influences production, and production influences perception, with substantial implications for social cognition.
    • This must be a foundational element of Social Trust. I see you do a thing. I imitate the thing. I feel (not think!) that it is the same thing. I do a thing. You imitate the thing. Think peekaboo. We establish a rapport. This is different from System Trust, where I put something somewhere and it’s still there. System trust may be derived fundamentally from Object Permanence, while Social Trust comes from imitation?
    • This is(?) tied to motor neurons. From Mirror neurons: Enigma of the metaphysical modular brain: Essentially, mirror neurons respond to actions that we observe in others. The interesting part is that mirror neurons fire in the same way when we actually recreate that action ourselves.
      • Implications for design? Journalism is definitely built around the ‘like me’ concept that it is built around stories. IR is much less so, and is more data focused.
    • At section 3 – Experiment 1: learning tool-use by observing others
      • We have Social Trust first. Then we learn to use tools. Tools are different from, though related to the environment. They are not ‘like me’, but they extend me (Heidegger again). More later.
  • Page 3: For example, there is an intimate relation between striving to achieve a goal and a concomitant facial expression and effortful bodily acts.
    • This is like the boot loader or initial dictionary entry. Hard-wired common vocabulary.
  • Page 3: Humans, including preverbal infants, imbue the acts of others with felt meaning not solely (or at first) through a formal process of step-by-step reasoning, but because the other is processed as ‘like me.’ This is underwritten by the way humans represent action—the supramodal action code—and self experience
    • So is there a ‘more like me’ and ‘less like me’?
  • Meeting with Wayne this evening
    • Go over notes
    • Coding session
  • ——————
  • Check to see that reports are being made correctly
    • Fix “Get all rated” Numerous issues, including strings with commas
    • Fix “Get Match Counts” all zeros
    • Fix “Get No Match Counts” redundent
    • Change “Get Blacklist (CSV)” to “Black/White list (CSV)
    • Add “Get Whitelist (Google CSE)
    • Change the Sets in getBlack/Whitelist to use maps rather than sets so blacklist culling can be used with more informative rows.
  • Update remote DB and test a few pages. Ran into a problem with LONGTEXT and Postgress. Went back to TEXT
  • Went over Aaron’s ASB slides a couple of times. Introduced him to Partial Least Squares Structural Equation Modeling (PLS-SEM).
  • Present new system to Andy, Margarita and John. Tomorrow…

Phil 4.13.16

7:00 – 4:30 VTX

  • One last thing from Deindividuation Effects on Group Polarization in Computer-Mediated Communication: The Role of Group Identification, Public-Self-Awareness, and Perceived Argument Quality. This is from the opening paragraph:
    • Group polarization refers to the well-established finding that following group discussion, individuals tend to endorse a more extreme position in the direction already favored by the group (Hogg, Turner, & Davidson, 1990; Isenberg, 1986; Moscovici & Zavalloni, 1969).
      • So this may require some expounding on, but it isn’t something that’s in dispute. Merging GPT with information network analytics in a way to simultaneously determine group membership while nudging for a view of a larger information horizon will require more scaffolding. But this plank is pretty solid.
      • And I like that my bibliography spans over 40 years and multiple disciplines.
  • I think I was able to put in a slot for the development of the slider functionality as it relates to a particular corpus. On a related note vector space classification is becoming a thing in NLP. In Deep or Shallow, NLP is Breaking Out in CACM, both Word2vec and GloVe are discussed. This ties back to The Hybrid Representation Model for Web Document Classification that I read back in January, where documents can be represented as clusters of vectors in an n-dimensional space. But now it looks like there are libraries. Woohoo! I do wonder if there’s a vector space analogy to chunking in this that could be useful. Maybe? Probably?
  • Starting The ‘like me’ framework for recognizing and becoming an intentional agent
    • Page 1: Autism has been described as a kind of ‘mind-blindness’ (Baron-Cohen, 1995) because children with autism do not conceptualize other people as psychological agents with a rich palette of mental states.
      • . What’s the influence on System Trust and Social Trust? Do groups of Autistic people polarize? Differently?
  • ————————-
  • Ok, after much flailing, I got the creation of queries correct for arbitrary PersonOfInterest. The problem is that beyond name and State, I really don’t know wht’s going to be used for a source file. So PoiObject has been generalized to handle name and state by default but everything else is optional. What was causing me all the trouble was having license_no as a ‘native’ value with a default of zero. When I made rules based on the presence or absence of this field, they couldn’t work. So much cleanup ensued…
  • Need to test the new data
    • Internal Exception: com.mysql.jdbc.MysqlDataTruncation: Data truncation: Data too long for column ‘source_url’ at row 1 – Fixed
      • Drop ‘queries’ out of provider menu dialog – Fixed
  • Need to change the queries in PoiObject that depended on licence_no. Commented out for now. Fixed.
  • Discussed geolocation with Aaron and sent out a note to the curation team for comments.

Phil 4.12.16

7:00 – 6:00 VTX

  • At the poster session yesterday, I had a nice chat with Yuanyuan about her poster on Supporting Common Ground Development in the Operation Room through Information Display Systems. It turns out that she is looking at information exchange patterns in groups independent of content, which is similar to what I’m looking at. We had a good discussion on group polarization and what might happen if misinformation was introduced into the OR. It turns out that this does happen – if the Attending Physician becomes convinced that, for example, all the instruments have been removed from the patient, the rest of the team can become convinced of this as well and self-reinforce the opinion.
  • Scanned through Deindividuation Effects on Group Polarization in Computer-Mediated Communication: The Role of Group Identification, Public-Self-Awareness, and Perceived Argument Quality. The upshot appears that individuation of participants acts as a drag on group polarization. So the more the information is personalized (and the more that the reader retains self awareness) the less the overall group polarization will move.
  • I’ve often said that humans innately communicate using stories and maps (Maps are comprehended at 3-4.5 years, Stories from when?). The above would support that stories are more effective ways of promoting ‘star’ information patterns. This is all starting to feel very fractal and self similar at differing scales…
  • Looking for children’s development of story comprehension led to this MIT PhD Thesis: TOWARD A MODEL OF CHILDREN’S STORY COMPREHENSION. Good lord – What a Committee: Marvin Minsky (thesis supervisor), Professors Joel Moses and Seymour Papert (thesis committee), Jeff Hill, Gerry Sussman, and Terry Winograd.
  • ———————
  • While reading Deep or Shallow, NLP is Breaking Out, I learned about word2vec. Googling led to Deeplearning4j.org, which has its own word2vec page, among a *lot* of other things. From their home page:
    • Deeplearning4j is the first commercial-grade, open-source, distributed deep-learning library written for Java and Scala. Integrated with Hadoop and Spark, DL4J is designed to be used in business environments, rather than as a research tool. Skymind is its commercial support arm.
    •  Deeplearning4j aims to be cutting-edge plug and play, more convention than configuration, which allows for fast prototyping for non-researchers. DL4J is customizable at scale. Released under the Apache 2.0 license, all derivatives of DL4J belong to their authors.
    •  By following the instructions on our Quick Start page, you can run your first examples of trained neural nets in minutes.
  • The word vector alternative is from the Stanford NLP folks: GloVe: Global Vectors for Word Representation. The link also has trained (extracted?) word vectors.
  • Testing the behavior of query construction and search results. Fixing stupid bugs. Testing more. Lathering, rinsing and repeating.
  • Some good discussions with Aaron on inferencing and toxicity profiles. Basically taking the outputs and determining correlations with the inputs. Which led to a very long day.

Phil 4.11.16

7:00 – 3:00 VTX

  • Make MOLST appointment today
  • Working on the outline
  • Continuing Technology, Humanness, and Trust: Rethinking Trust in Technology. Done! Meaty.
    • Page 907: While the relationship between human-like trust and outcomes is stronger in most cases, system-like trust still matters (see Figure 2). This may be because, in part, Facebook is a tool that helps one do social networking. I think this is important. When a tool has high system trust in a social context, it disappears, while the social aspect comes to the fore. This is true even if the tool is performing hidden tasks that influence the social interaction. This is related to relevance and pertinence, I think. As long as the social cies are presented in a way that feels pertinent (and is reliable?), it’s trusted explicitly as a system and implicitly as a player in the social interaction.
    • Clifford Nass, Byron Reeves
  • ——————————-
  • TODOs in GoogleCSE2 buildQueryObjects() and buildNewQueryObjects():
    • Modify to take SmartTerm Object
      If there is a valid term, then create the query with po.getNamePermutations()
  • Finished. Building a new (small!) set of people to test with
  • Discussed how order affects search results with Andy. Need to think about that
  • Had an idea about running overspecified queries that return nothing, then backing off term by term until a hit. Running through the permutations that have very small numbers of hits looking for common hits might be a good way of getting good results?

Phil 4.8.16

7:00 – 4:30 VTX

  • Here’s the new link for Microsoft Cognitive Services
  • Continuing Technology, Humanness, and Trust: Rethinking Trust in Technology.
    • Page 906: Among the five factors, social presence correlated the highest with humanness for both Facebook (0.48) and Access (0.56). Also noteworthy is that for Access, the correlation between humanness and animation was high (0.51), whereas for Facebook it was not (0.31). Further, dynamism correlated somewhat higher with humanness for Access (0.39) than for Facebook (0.23). These interesting differences show that each technology likely has a general humanness that finds its basis in different factors
      • This leads me to believe that ‘humanness’ is not exactly what they are testing here. Responsiveness can be used to discriminate between different types of tires (WRT cornering), and I don’t think anyone would call one tire more or less human than another. I think this also applies to the animation test. Social presence though makes a lot of sense.
      • It did just strike me that partial least squares structural equation modeling (PLS-SEM – [XLstat’s definition and tutorial) would be a *great* way of evaluating trustworthiness and credibility cues. This should be part of the research part of the study?
    • Page 906: Our study 2 findings raise a related question: instead of considering humanness a general construct measured with three items like we did, could one theorize humanness as a second-order construct that is reflected by specific first-order factors that are components of social presence, social affordances,and affordances for sociality? Researchers exploring such a second-order construct could integrate it into a nomological humanness network.
      • Hah! See my second comment for the previous quotation.
    • Page 907: Researchers should try to determine differences between these respondents and those who ranked one or both technologies at or above the midpoint. Researchers could also perform a cluster analysis to identify groups with common responses to the humanness items of which a group with low humanness scores might emerge. It could be that the humanness factors identified in study 2 are more or less important in indicating humanness based on cluster membership. It could also be that results from study 1 about the importance of trust type might differ by humanness cluster.
      • This could also be a component of trust/credibility analysis.
    • Paper structure thoughts
  • —————————
  • Set the persistence.xml to point to the Talend DB
  • Created the DB
  • Added users
    • Phil
    • Aaron
    • Margarita
    • Andy
    • John
  • Need to figure out how to come up with a list of names/terms/CSEs to start evaluating
  • Need to test fully functional app, then package and deploy
  • Need to have VTX get a SemRush account
  • Conference Call with John, Margarita and Andy about setting up the Crawl for this weekend. John will get back to me with some known bad actors
  • Need to associate search terms with an optionalString element
  • Monday’s TODOs in GoogleCSE2 buildQueryObjects() and buildNewQueryObjects():
    • Modify to take SmartTerm Object
      If there is a valid term, then create the query with po.getNamePermutations()

Phil 4.7.16

From Communications of the ACM’s Kode Vicious column: To understand the first downside, you should find a friend who works on compilers and ask if he or she has ever looked inside gcc (GNU C compiler), and, after the crying stops and you have bolstered your friend’s spirits, ask if he or she has ever tried to extend the compiler. If you are still friends at that point, your final question should be about submitting patches upstream into this supposedly open source project.

Yup.

7:00 – 4:30 VTX

  • Continuing Technology, Humanness, and Trust: Rethinking Trust in Technology.
    • Page 894: trust is most often treated as a psychological construct (i.e., trusting
      beliefs). As a psychological construct, trusting beliefs exists apart from any attempt to measure it (Schwab, 1980). Yet knowing what the construct means helps one to measure it properly. Hence, the trusting beliefs construct will influence its components. Third, we used reflective first-order factors because we did not seek to explain variance in trusting beliefs.

      • So trust can be measured using inferential models? As an influence system maybe???
    • At 6.2. Study 2: Methodology, page 903. The second study is more related to the credibility cues that people use to determine the humanness of an interface. Not sure if it’s relevant to what I’m working on, but it is interesting to see how they include the second study which follows up on the open questions from the first.
  • In the paper above, they use something called partial least squares structural equation modeling (PLS-SEM). SmartPLS is a system that uses this, and there’s a presentation on YouTube that shows how it’s used to predict shadow banking. Need to look into this some more as a way of predicting outcomes based on behavior.
  • ———————–
  • Sent an email to John and Bob about using the new CSEs
  • Set up the rating app so that Andy and Margarita can use it to create the json characterization. Had a hell of a time getting the executable jar built. The artifact builder in Intellij doesn’t synchronize with the dev process. I was not including jars that were required and getting a “Error: A JNI error has occurred, please check your installation and try again” error on execution. I wound up having to delete the artifact, commit, create a new artifact and then create the jar and executable.
  • Sent the Rating app as a zip. Not sure if the filters are letting it through. Hey! It works!
  • Sent Aaron a rant on what I’d like to get the db up and running. Done! Yay!
  • Finalized REST discussions with Jeremy