Latent Semantic Analysis Similarity Calculator

This container is used to perform Latent Semantic Analysis (LSA) on short texts. This includes the ability to create a semantic space as well as the ability to calculate cosine similarity scores on short texts using a semantic space.

Running the Tool

In order to run the tool, type the following into your terminal

docker run -it -v /local/data/path/:/app/data -e CONFIG_FILE=config.yml o76923/lsa

Where /local/data/path/ is the path on your local system that contains the source information and config.yml is the configuration file in that directory containing which tasks should be performed as well as their settings.

Directory Structure

The path on your local machine should contain the following

1. A configuration file
2. The semantic space
3. The texts to be compared

Configuration File

The configuration file specifies the parameters that will tweak how the tool behaves. A sample configuration file is provided in /app/conf/config.yml. The sections of it are as follows

Tasks

There are two main tasks that can be performed by this tool: "create_space" and "calculate_similarity".

create_space

The create_space task is used to create the semantic space that will be used from a source paragraph given a few settings. A sample create_space task is included below, followed by an explanation of the options available.

  - type: create_space
    space: PR
    space_settings:
      stem: false
      case_sensitive: false
      dimensions: 500
      remove:
        - punctuation
        - singletons
        - numbers
        - stopwords:
            library: nltk
    from:
      document_scope: line
      files:
        - paragraphs/PR.txt

space

The name that you wish to give the newly created semantic space.

stem

Should the words be stemmed using the Porter stemmer?

case_sensitive

Should words be converted to lower case before processing?

dimensions

What should the rank of the vectors created in the space be?

remove

Should punctuation, singletons, or numbers be removed? If present in the list, they will be removed; otherwise they are retained).

stopwords

If stopwords are removed, where should the list of stopword come from? At this time, the only option supported is to note that the stopwords list from nltk should be used.

document_scope

What defines a document for purposes of reading in source files? At the moment, only "line" is supported meaning each line of the file is treated as a different document.

files

A list of files that contain the source documents that you want used in the semantic space.

calculate_similarity

The calculate_similarity task is used to generate semantic similarity scores between short texts. A sample calculate_similarity task is included below, followed by an explanation of the options available.

  - type: calculate_similarity
    options:
      distance_metric: cosine
      space: Bus
    from:
      files:
        - input/name.txt
      pairs: all
      headers: true
      numbered: true
    output:
      format: H5
      file_name: name.h5
      ds_name: lsa_bus

space

The name of the semantic space to be used when calculating similarities.

distance_metric

The metric used when comparing similarities. Options are either cosine or r.

files

A list of files that contain the short texts to be compared.

pairs

Which pairs of short texts should be compared to one another? At the moment, "all" is the only option supported which compares each text to every other text.

headers

Do your files contain a header row that should be skipped?

numbered

Do the texts have IDs assigned to them already?

format

What format should the similarity scores be written to? At the moment, only "H5" is available which saves the output in the HDF5 file format.

file_name

The name of the similarity file. It will be placed in the `/app/data/output` directory.

ds_name

If the format is "H5", you can specify the name of the data source. This name will be used in both the sims and vector groups.

Options

Options specifies global options that will apply to all tasks run. At this time, only one option is available.

cores

The number of processor cores that can be used at any given time.

Semantic Space

The semantic space is the corpus that is used in order to create similarity files. It is the output from a "create_space" task.

Texts to be Compared

The texts to be compared are the short texts that you wish to have compared to one another. Similarity scores will be generated between texts with one ID and texts with another ID.