Major Documentation Revamp

[Ayushk4]

This PR is aimed at improving the docs in the following ways -

• Adding examples
• Make it easy to understand
• Improving the styling
• Fixing typos and grammatical errors

Also testing for the possible errors in the package.

Progress

• Documents
  • Creating documents
  • Basic functions for working with Documents
  • Document Metadata
  • Preprocessing
• Corpus
  • Creating a Corpus
  • Standardizing a Corpus
  • Processing a Corpus
  • Corpus Statistics
  • Converting a DataFrame from a Corpus
  • Corpus Metadata
• Features
  • Creating a Document Term Matrix
  • Creating Individual Rows of a DTM
  • Hash Trick
  • TF-IDF
• Semantic
  • LSA
  • LDA

[Ayushk4]

I am also replacing the deprecated functions in the documentation. So, far I have finished documents.md.

[Ayushk4]

@aviks I was going through the docs. I found that names (used for setting metadata - author entire corpus of docs name for the docs in the corpus ), conflicts with the base refer (v0.7) and (v1). What do you suggest about this? Should I open an issue?

[aviks]

`author` maybe?

…

On Sun, Mar 17, 2019 at 9:06 AM Ayush Kaushal ***@***.***> wrote: @aviks <https://github.com/aviks> I was going through the docs. I found that names <https://github.com/JuliaText/TextAnalysis.jl/blob/b3c467cebb776c6a7e686cfe2b6c9a20c2ca2205/src/metadata.jl#L32> (used for setting metadata - author name for entire corpus of docs), conflicts with the base refer (v0.7) <https://docs.julialang.org/en/v0.7/base/base/#Base.names> and (v1) <https://docs.julialang.org/en/v1/base/base/#Base.names>. What do you suggest about this? Should I open an issue? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#134 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAXIJuB1qmtFTXVLGmXZV1DwMdYi5POJks5vXgWRgaJpZM4b3U1e> .

[Ayushk4]

A correction - Author/authors is already used for document's author, should I change the metadata containing the document name to documentName. Also, I am sending a separate PR for this.

[Ayushk4]

The following are the metadata - names -> setting the document's name and authors for setting the author's name.

This is also missing from the documentation for accessing the corpus metadata, I am adding it.

[zgornel]

• Documenter.jl seems to be at this point a better choice since all code snippets are (re-)evaluated at branch modifications and reflect changes in the code. Is there any reason for not using it ?

• A more flexible approach to the metadata field names is to allow a metadata type hierarchy (i.e. define an AbstractMetadata type) https://zgornel.github.io/StringAnalysis.jl/dev/doc_extensions/

[Ayushk4]

• I don't see any downside of using Documenter.jl features of @repl and doctests. Maybe this can be reflected across the documentation of JuliaText Repositories.

• A lot of things has to be ported from StringAnalysis.jl. I came across this issue - Merge back to TextAnalysis zgornel/StringAnalysis.jl#2 (comment).

[Ayushk4]

I think that maybe we can start by opening an issue of what all major changes are to be done for TextAnalysis.jl, taking into consideration the things mentioned in the conversation above.
Thank you, for your review on the situation.

[Ayushk4]

@aviks Can you please suggest your views on this, so that I may proceed accordingly.

[aviks]

Thanks Ayush, this looks good. Let me know once this is ready to merge. I'm happy to merge things over from StringAnalysis if Corneliu is ok with that.

[zgornel]

@aviks I'm perfectly fine with this; feel free to port to TextAnalysis whatever you see fit

[aviks]

Separate one please

…

On Mon, 18 Mar 2019, 18:48 Ayush Kaushal, ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In docs/src/corpus.md <#134 (comment)> : > @@ -83,3 +139,56 @@ corpus. The easiest way to do this is to convert a `Corpus` object into a `DataFrame`: convert(DataFrame, crps) + +## Corpus Metadata + +You can also retrieve the metadata for every document in a `Corpus` at once: + +* `languages()`: What language is the document in? Defaults to `Languages.English()`, a Language instance defined by the Languages package. +* `names()`: What is the name of the document? Defaults to `"Unnamed Document"`. Do you prefer that I send another PR making changes, name -> title & name! -> title! & names -> titles & names! to titles! (across the entire codebase, including tests, docs) ? Or include in this one itself? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#134 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAXIJswvIK-wRXsrqwkjVruj7A05Stdaks5vX9-NgaJpZM4b3U1e> .

[Ayushk4]

@aviks I am almost done with the PR. All that remains to be done, is #135 being merged so that I can remove the merge conflicts that may arise.

Also, lsa() is not working. I will get more details about it and file a detailed issue.

[Ayushk4]

@aviks Do you think, BM25 and Co-occurrence matrix will be a good addition to TextAnalysis.jl?

[aviks]

@aviks Do you think, BM25 and Co-occurrence matrix will be a good addition to TextAnalysis.jl?

Yes please!