Data Preparation

This package deals with short text. While the text data for predictions or classifications are simply str or list of str, the training data does take a specific format, in terms of dict, the Python dictionary (or hash map). The package provides two sets of data as an example.

Example Training Data 1: Subject Keywords

The first example dataset is about the subject keywords, which can be loaded by:

>>> trainclassdict = shorttext.data.subjectkeywords()

This returns a dictionary, with keys being the label and the values being lists of the subject keywords, as below:

{'mathematics': ['linear algebra', 'topology', 'algebra', 'calculus',
  'variational calculus', 'functional field', 'real analysis', 'complex analysis',
  'differential equation', 'statistics', 'statistical optimization', 'probability',
  'stochastic calculus', 'numerical analysis', 'differential geometry'],
 'physics': ['renormalization', 'classical mechanics', 'quantum mechanics',
  'statistical mechanics', 'functional field', 'path integral',
  'quantum field theory', 'electrodynamics', 'condensed matter',
  'particle physics', 'topological solitons', 'astrophysics',
  'spontaneous symmetry breaking', 'atomic molecular and optical physics',
  'quantum chaos'],
 'theology': ['divine providence', 'soteriology', 'anthropology', 'pneumatology', 'Christology',
  'Holy Trinity', 'eschatology', 'scripture', 'ecclesiology', 'predestination',
  'divine degree', 'creedal confessionalism', 'scholasticism', 'prayer', 'eucharist']}

Example Training Data 2: NIH RePORT

The second example dataset is from NIH RePORT (Research Portfolio Online Reporting Tools). The data can be downloaded from its ExPORTER page. The current data in this package was directly adapted from Thomas Jones’ textMineR R package.

Enter:

>>> trainclassdict = shorttext.data.nihreports()

Upon the installation of the package, the NIH RePORT data are still not installed. But the first time it was ran, it will be downloaded from the Internet.

This will output a similar dictionary with FUNDING_IC (Institutes and Centers in NIH)

as the class labels, and PROJECT_TITLE (title of the funded projects)

as the short texts under the corresponding labels. This dictionary has 512 projects in total, randomly drawn from the original data.

However, there are other configurations:

Example Training Data 3: Inaugural Addresses

This contains all the Inaugural Addresses of all the Presidents of the United States, from George Washington to Barack Obama. Upon the installation of the package, the Inaugural Addresses data are still not installed. But the first time it was ran, it will be downloaded from the Internet.

The addresses are available publicly, and I extracted them from nltk package.

Enter:

>>> trainclassdict = shorttext.data.inaugural()

User-Provided Training Data

Users can provide their own training data. If it is already in JSON format, it can be loaded easily with standard Python’s json package, or by calling:

>>> trainclassdict = shorttext.data.retrieve_jsondata_as_dict('/path/to/file.json')

However, if it is in CSV format, it has to obey the rules:

  • there is a heading; and

  • there are at least two columns: first the labels, and second the short text under the labels (everything being the second column will be neglected).

An excerpt of this type of data is as follow:

subject,content
mathematics,linear algebra
mathematics,topology
mathematics,algebra
...
physics,spontaneous symmetry breaking
physics,atomic molecular and optical physics
physics,quantum chaos
...
theology,divine providence
theology,soteriology
theology,anthropology

To load this data file, just enter:

>>> trainclassdict = shorttext.data.retrieve_csvdata_as_dict('/path/to/file.csv')

Home: Homepage of shorttext