All Projects → lambdahands → lingo

lambdahands / lingo

Licence: EPL-1.0 license
A Clojure natural language generator built on top of University of Aberdeen’s SimpleNLG library.

Programming Languages

clojure
4091 projects

lingo

A Clojure natural language generator built on top of University of Aberdeen’s SimpleNLG library.

Usage

  1. Clone the repository to your local file system.
  2. Run lein repl in the lingo directory.
  3. Do: (use 'lingo.core)
  4. And: (use 'lingo.features)
  5. Hack away!

Todo

  • Test custom lexicons.
  • Integrate core.logic in some bizarre and amazing way (meters anyone?)
  • Much, much more...

Walkthrough

This walkthrough is available as a Clojure file in the examples directory. It's encouraged that you go through it interactively.

Lingo's pieces comprise Clojure data structures. Maps are the primary structure used to express each part of speech we're interested in.

For example, an empty clause looks like this:

{:> :clause}

The key :> is the identifier key. It tells us what kind of part we plan on generating. Here are other parts we're able to generate:

{:> :noun}
{:> :verb}
{:> :subject}
{:> :object}

Generating any of these parts will create empty phrases. To do something more useful, we're able to "add" specific pieces to our parts.

A noun can simply pass a string or a vector of two strings, the first being what's known as the determiner ("the", "a", or "that") and the second being the actual phrase.

{:> :noun :+ "dog"}
{:> :noun :+ ["the" "dog"]}

However verbs accept only one string.

{:> :verb :+ "run"}

Subjects and objects are only nouns, but they're used to specifically create clauses. Clauses take a vector of one of each a subject, object and verb as well as many complements as you wish. We'll cover complements soon.

(def dog-and-rabbit
  {:> :clause
   :+ [{:> :subject :+ ["the" "dog"]}
       {:> :verb    :+ "chase"}
       {:> :object  :+ ["the" "rabbit"]}]})

In order to generate our pieces into speech, we need to create a generator.

(def generator (make-gen))

The generator contains two primary functions, accessed as keys to a Clojure map.

The first creates SimpleNLG objects. Once created, you can use Java methods to access fields and (not recommended!) mutate them.

(:* generator)

The second realises- or "renders"- our speech into a sentence. It's recommended to do as many operations on Clojure data structures before passing them to the generator.

(:! generator)

We pass the realisation function to our dog and rabbit clause to see the generated result.

((:! generator) dog-and-rabbit)
;; => "The dog chases the rabbit."

We can manipulate our clauses just as maps, and we can even make philosphical inqueries.

((:! generator)
 (assoc dog-and-rabbit :* {:feature [:why :?]}))
;; => "Why does the dog chase the rabbit?"

By assoc-ing the :* key with a feature, we were able to turn our statement into a question. The :* key is known as the modifier key.

Modifiers can be a great deal of different things. Here are several examples of what you can pass as the :* key:

{:> :verb :+ "run" :* "quickly"} ;; Add an adverb
{:> :noun :+ "dog" :* "confused"} ;; Add an adjective
{:> :noun
 :+ "management"
 :* [:pre "time"]} ;; A pre-modifier resides before the phrase.
{:> :noun
 :+ "management"
 :* [:post "time"]} ;; A post-modifier resides after the phrase.

Complements are similar to modifiers, but they can represent phrases in a general sense, from simple adverbs to prepositional phrases. In linguistics, a complement is loosely defined as anything that comes after the verb.

(def park-chase
  (assoc dog-and-rabbit :* {:complement "around the park"}))
((:! generator) park-chase)
;; => "The dog chases the rabbit around the park."

We can add as many features and complements, in any order, as we like.

((:! generator)
 (merge
   dog-and-rabbit
   {:* [{:feature [:how :?]}
        {:feature [:is :perfect]}
        {:feature [:future :tense]}
        {:complement "around the park"}]}))
;; => "How will the dog have chased the rabbit around the park?"

Modifiers can even be other parts to create coordinated phrases.

((:! generator)
 {:> :noun :+ "Jack"
  :* [{:> :noun :+ "Wendy"}
      {:> :noun :+ "Danny"}]})
;; => "Jack, Wendy and Danny."

We can combine all of these pieces to create well known phrases.

(def redrum
  {:> :clause
   :+ [{:> :subject
        :+ "work"
        :* [{:> :noun :+ "play" :* [:pre "no"]}
            [:pre "all"]]}
       {:> :verb :+ "make"}
       {:> :object :+ "Jack"}
       {:> :complement :+ "a dull boy"}]
   :* {:feature [:base-infinitive :form]}})

((:! generator) redrum)
;; => "All work and no play makes Jack a dull boy."
Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].