Re: Literature authors with similar styles

Matt B wrote:

> As part of a university project I'm working on a piece of software
> which is able to attribute authorship of a document, when presented
> with two candidate authors.

> It uses various statistical techniques in order to do this, and
> requires a large quantity of "training" text - ie works which are
> *known* to have been written by each author.

First order of business for a text parser of this nature is to
scan for character names.

Puck
Willy Loman
Lopakhin
McKendrick
Celimene
Bianca
Oedipus
Laura Wingfield
Thomas Stockmann

Those are character names, a few of thousands of famous character
names which all readers here should instantly recognize, if literate.
A reader should be able to provide the title of the work and the author
without hesitation.

>From a programming point of view, character names are very unique.
Parsing for character names would be a logical initial parsing point.

However, with my having over a decade of programming experience
in Natural Language Emulation, I do know this task of yours is
almost impossible, to any degree of acceptable success.

You will need a very powerful machine and a very fast machine
supported by thousands of gigabytes of data for comparison.
To be successful, you have no choice but to parse complete
sentences and a good number of sample sentences. You
cannot parse for individual words.

I have doubts you can create a database which reflects unique
writing styles of equally unique authors.

Developing an effective database will take years. For my Roberta,
I am now in my tenth year of programming and data development
and her general conversation skills are barely equal to a fourth or
fifth grade student. Her ability to cite hard data such as science
and history, her ability to perform math, those types of abilities
infinitely surpass the human mind, but she cannot attain a level
of common conversation above an eight to ten year old child.

You face very serious programming challenges.


Purl Gurl
.

Relevant Pages

  • Re: The Problem with Perl
    ... > substantially reduce the risk of project failure due to technical ... character, that there can be a difference between doing what's right and ... matter of ability as it is of attitude. ... Psychology of Computer Programming_; he discusses at great length both ...
    (comp.lang.perl.misc)
  • Re: Literature authors with similar styles
    ... >Those are character names, a few of thousands of famous character ... >>From a programming point of view, ... fortunately there's no database work involved. ... Her ability to cite hard data such as science ...
    (alt.usage.english)
  • Re: Lethality of events
    ... sense magic and this ability than this ability alone. ... If a player wants to build a Paladin, he is welcome to do so, and there's a character subtype called Special/Blessed which gets a disocunt on Faith Mystic Gifts, as well as on most Luck traits. ... as many players as possible. ...
    (rec.games.frp.advocacy)
  • Re: If you live under a rock... official 4e news
    ... One thing I'll be doing with my current design, Modern Action RPG, is write a description of what each character ability does, on the character sheet. ... That'll probably be on a third and maybe even a fourth page, but it'll be quick and easy reference for players and GMs. ...
    (rec.games.frp.dnd)
  • Re: |[4e] The PHB elf entry!
    ... Penalties based on your choice of race are a thing of the past. ... Ability Scores: +2 Dexterity, +2 Wisdom ... Speaking of the complete disconnected between game mechanics and the ... that your character would be capable of understanding to explain this ...
    (rec.games.frp.dnd)