subject-levellers-freebo

This is an automatically generated overview of the Distant Reader study carrel called subject-levellers-freebo.

Given a corpus of narrative text, the Distant Reader and the Distant Reader Toolbox create data sets, and these data sets are affectionately called "study carrels". As data sets, study carrels are intended to be read by people as well as computers, and their purpose is to supplement the reading process, to increase use & understanding.

Study carrels are designed to be computable, and this Web page is the result of one such computing process; here you will find a simple analysis of the carrel's extracted features. Use the features to characterize the content of the carrel, and then use them like items in a back-of-the-book index for more in-depth analysis.

Basic characteristics

Outlined below are some introductory features of the carrel:

Sizes

Measured in words, how big are the items in this carrel? To what degree are they of similar size? Are any of the items particularly large or small? In general, the typical scholarly journal article is between 5,000 and 8,000 words long.

Readability

Using a metric called Flesch -- where 0 means nobody can read the item, and 100 means everybody can read it -- how difficult are the items in this carrel to read? Shakespeare's Sonnets are relatively easy to read (with a score of about 90) because their vocabulary is small and the sentences are short. The typical novel by Jane Austen or Herman Melville have scores in the 70's. Scholarly articles are more difficult (scores between 50-70) because their langauge is more specialized. Texts of poor optical character recognition quality typically have very low scores.

Ngrams

Excluding stop words, what are the most frequent individual words and two-word combinations in the corpus, thus addressing the question, "What are the items in this study carrel about?"

unigrams

bigrams

Parts-of-speech

The most frequent extracted parts-of-speech features address questions of "What is discussed in this corpus, what do they do, and how are they described?"

nouns	proper nouns
pronouns	verbs
adjectives	adverbs

Entities

Similar to nouns, named-entities are real-world things but they are more specific. They help address questions of who, where, and how many.

any entity	persons
geo-political entities	organizations

Clusters

Through the use of a variation of the principle component analysis alogorithm, it is possible to plot the location of study carrel items in two- and three-dimentional spaces, thus addressing the questions, "To what degree is this study carrel holistic; to what degree are the items in this carrel easily subdivided into smaller group?"

Keywords

Excluding stop words, and through the use of a variation of the term frequency-inverse document frequency algorithm, the set of computed keywords are akin to subject terms, and they help address the questions of, "What are the items in this carrel about and to what degree?"

Additional models

Depending on how the carrel was computed against, the following point to additional models:

• analysis - if a person used the modeling tools to analyze this carrel and wrote up their observations, then those observations ought to be here
• bibliographics (txt) - authors, titles, dates, extents, summaries, and keywords in a simple human-readable form
• bibliographics (JSON) - same as the above but formatted as a JSON stream
• compressed- the whole study carrel compressed into a single file for the purposes of collaboration, sharing, and downloading
• manifest - a browsable interface to the study carrel
• metadata - if the study carrel creation process was augmented with metadata values (authors, titles, dates, etc.) file, then that file is available here
• network graph - a Graph Modeling Language file of the carrel's author(s), titles, and computed keywords, and it is useful for visualizing their relationships
• provenance - a very very rudimentary list of characateristics denoting whence the carrel came and when
• semantic triples - bibliographic characteristics encoded in the form of the Resource Description Framework, and intended for the purposes of supporting the Semantic Web
• summary - this file

For more detail about study carrels, their structure, and how they can be used, start with the read me file.

Subdirectories

Each and every study carrel contains a number of subdirectories (folders). The first two are about the carrel's content:

• cache - here you will find the original content used to create the carrel
• txt - this subdirectory contains plain text versions of original content; all analysis is done against the content in this directory

The next few subdirectories contain extracted features in the form of tab-delimited text files:

• adr - email addresses, if they exist
• bib - bibliographics (authors, titles, dates, extents, and summaries)
• ent - named-entities (real world things such as people, places, organizations, dates, times, etc.)
• pos - parts-of-speech (each and every word from each and every item described as a noun, verb, adjective, etc.)
• urls - Universal Resource Locators, if they exist
• wrd - statistically significant computed keywords

Very important. All of the content in the subdirectories above are readable by any spreadsheet application, database program, or programming language. Therefore, you do not need special software to do analysis.

There are two additional subdirectories in every study carrel:

• figures - where images are saved
• etc - the subdirectory for everything else, and of greatest importance is the carrel's stop word list, bag-of-word representation of the carrel, and the carrel's SQLite database file

All of this is just the beginning. For more detail about study carrels, their structure, and how they can be used, begin with the read me file.

Date created: 2024-10-13