Logo
  • Installation
    • Recommended Install
    • Full Install
      • Setup
        • Linux
        • OSX (with brew)
        • Install (pip)
        • Install (Anaconda)
        • Install (GitHub)
      • Install Other Dependencies
        • Load NLTK corpus
        • Install Stanford Parser
        • Install Tweet NLP v0.3 (ark-tweet-nlp-0.3)
        • Python Modules (optional)
        • Install the IBM Wordcloud jar file (optional)
        • Mallet (optional)
    • Full List of Dependencies
      • Python
      • Other
      • Python (optional)
      • Other (optional)
    • Python version support
    • Getting Started
      • Command Line Interface
      • MySQL Configuration
      • Sample Datasets
      • Next Steps
    • Install Issues
  • Github Repo
  • Getting started in Colab
  • Tutorials
    • Getting started
    • Text Cleaning and Transformations
    • Feature Extraction
    • Viewing your data and output
    • Prediction
    • Clustering
    • LDA with Mallet
    • Data Engines
    • Other Topics
    • Video Tutorials
  • Packaged Datasets
    • Language Data
      • Blog Authorship Corpus
    • Lexica
      • Age and Gender Lexica
      • PERMA Lexicon
      • Spanish PERMA Lexicon
      • Other Lexica
    • LDA Topics
      • 2000 Facebook Topics
  • dlatkInterface Flags by type
    • Setup
    • Preprocessing
    • Feature Extraction
    • Feature Refinement
    • Language Insights
    • Clustering
    • Prediction
      • Regression
      • Classification
    • Visualization
  • Papers Utilizing DLATK
    • DLATK Paper
    • Peer Reviewed Publications
      • 2020
      • 2019
      • 2018
      • 2017
      • 2016
      • 2015
      • 2014
      • 2013
DLATK
  • --print_joined_feature_lines
  • View page source

--print_joined_feature_lines

Switch

--print_joined_feature_lines

Description

Accomplishes something similar to --print_tokenized_lines, but starting from a feature table instead of a message table. Result is formatted for input to Mallet; see LDA Tutorial.

Argument and Default Value

If the feature is a multi-word expression joined by spaces, this will replace the spaces with underscores.

Details

This is useful for creating topics from collocations without turning the collocation table back into a message table.

If value > 1, prints that feature [value] times.

If this is run on a non:doc:fwflag_occurence-filtered 1:doc:fwflag_gram table grouped on message_id, this will act identically to --print_tokenized_lines (though order of tokens may be different).

Other Switches

Required Switches: -d, -t, -f Optional Switches: None

Example Commands

Input feature table:

+----+----------+---------------+-------+--------------------+
| id | group_id | feat          | value | group_norm         |
+----+----------+---------------+-------+--------------------+
|  1 |        1 | a             |     1 | 0.0714285714285714 |
|  2 |        1 | friend        |     1 | 0.0714285714285714 |
|  3 |        1 | off           |     1 | 0.0714285714285714 |
|  4 |        1 | very          |     1 | 0.0714285714285714 |
|  5 |        1 | =)            |     1 | 0.0714285714285714 |
|  6 |        1 | is            |     1 | 0.0714285714285714 |
|  7 |        1 | well          |     1 | 0.0714285714285714 |
|  8 |        1 | .             |     2 |  0.142857142857143 |
|  9 |        1 | to            |     1 | 0.0714285714285714 |
| 10 |        1 | rested        |     1 | 0.0714285714285714 |
| 11 |        1 | starbucks     |     1 | 0.0714285714285714 |
| 12 |        1 | with          |     1 | 0.0714285714285714 |
| 13 |        1 | to catch up   |     1 | 0.0714285714285714 |
| 14 |        2 | and           |     2 | 0.0666666666666667 |
| 15 |        2 | city          |     1 | 0.0333333333333333 |
| 16 |        2 | "             |     4 |  0.133333333333333 |
| 17 |        2 | two           |     1 | 0.0333333333333333 |
| 18 |        2 | .             |     2 | 0.0666666666666667 |
| 19 |        2 | to            |     1 | 0.0333333333333333 |
| 20 |        2 | memory        |     1 | 0.0333333333333333 |
| 21 |        2 | new           |     1 | 0.0333333333333333 |
| 22 |        2 | :             |     1 | 0.0333333333333333 |
| 23 |        2 | excited       |     1 | 0.0333333333333333 |
| 24 |        2 | today         |     1 | 0.0333333333333333 |
| 25 |        2 | soundtrack's  |     1 | 0.0333333333333333 |
| 26 |        2 | got           |     1 | 0.0333333333333333 |
| 27 |        2 | oddessy       |     1 | 0.0333333333333333 |
| 28 |        2 | albums        |     1 | 0.0333333333333333 |
| 29 |        2 | by            |     1 | 0.0333333333333333 |
| 30 |        2 | i am very     |     1 | 0.0333333333333333 |
| 31 |        2 | this          |     1 | 0.0333333333333333 |
| 32 |        2 | motion        |     1 | 0.0333333333333333 |
| 33 |        2 | zombies       |     1 | 0.0333333333333333 |
| 34 |        2 | oracle        |     1 | 0.0333333333333333 |
| 35 |        2 | commit        |     1 | 0.0333333333333333 |
| 36 |        2 | the           |     3 |                0.1 |

Print command, using the packaged dla_tutorial dataset (see Differential Language Analysis (DLA) Tutorial), grouped at the message level:

dlatkInterface.py -d dla_tutorial -t msgs -c message_id -f 'feat$colloc$msgs$message_id$16to16' --print_joined_feature_lines featCollocLines.txt

Produces in featCollocLines.txt:

1 en a off very with is well . . to rested starbucks to_catch_up =) friend
2 en and and city " " " " two . . to memory got : excited today soundtrack's new
   oddessy albums by i_am_very this motion zombies oracle commit the the the

© Copyright 2024, H. Andrew Schwartz and Salvatore Giorgi.

Built with Sphinx using a theme provided by Read the Docs.