agoldst / dfr-browser Goto Github PK

My first attempt works fine in Firefox, but seems not to load the data/models in Safari (Version 9.1 (11601.5.17.1)) or Chrome (Version 51.0.2704.19 dev (64-bit)). This happens when I Open the file in the browser, set up a local server, or try a server on the internet.

I'm working on adapting the topic modeling tool to work with this wonderful piece of software. In the process of getting a testing workflow set up, I found that the standard citations.tsv file provided by the DFR tool, when passed to prepare-data, appears to generate a rogue line like this at the end:

"10.2307/461391","The Telltale Teeth: [...]
"",""                                         < -- here lieth a TypeError

JavaScript becomes confused when it tries to read the data from this line, and dies.

I think this is caused by an unfortunate blank line at the end of citations.tsv. Fixing should be as simple as removing any blank lines from the input before processing, but I wanted to pass it by you before submitting a PR.

Just putting a marker here for a significant rewrite for the next version of dfr-browser, which allows for swapping among models or metadata covariates. (Visualizing more than one model / covariate at once is a yet later revision). I have been sitting on working code for this for a while, but I don't want to release it until the companion data-export functions for dfrtopics are ready.

Hi there,

1st, thank you so much for the project 👍

In the master version, when I run it, under topic view, the x-axis of "yearly proportion of words in topic" plot seems to be "covered" and went missing.

I'm using Chrome/Ubuntu.

I built a topic model and saved the files (.gz, .mallet) to load into dfr-browser. What is the metadata_file="dfr-data/citations.tsvfile and what format should it follow?

I switch to the gh-pages branch and run the demo/index.html file. It says:

Unable to load topic words from data/tw.json
Unable to load model info from data/info.json

Do you know why is that?

First of all kudos 💯 for this amazing data explorer! I know that this question has been asked several times, my question is how to start with that in the case I have a very basic/raw LDA implementation (LDA/Gibbs) where the output data are like:

{
    "topic": 1,
    "terms": [
      {
        "term": "love",
        "weight": 0.068
      },
      {
        "term": "body",
        "weight": 0.03
      },
      {
        "term": "baby",
        "weight": 0.027
      },
      {
        "term": "girl",
        "weight": 0.025
      },
      {
        "term": "dance",
        "weight": 0.02
      }
    ]
  }

i.e. the array of topics plus the metadata of the referring documents of course.

In recent versions of WebKit (Safari 9.1, Chrome 50.0.2661.86) declaring use strict in utils.min.js seems to lead to the following error

TypeError: Attempted to assign to readonly property.

I believe the line in the minified version:

f[r] = t(n[r], e[r])

corresponds with line 58 of the non-minified version:

result[prop] = deep_replace(x[prop], repl[prop]);

It's entirely possible this is a WebKit bug; in the meantime, removing

"use strict";

from utils.min.js seems to make the problem go away.