prot.clj

(ns clomoios.contextsearcher
  (:use [clomoios.core])
  (:require [opennlp.nlp :as nlp]))

(defprotocol Searcher
  "An interface for searching"
  (score  [this term text] "Score this text in similarity")
  (rank   [this term text] "Rank sentences in this text"))


(defrecord ContextSearcher [get-sentences tokenize pos-tag])

(extend-protocol Searcher ContextSearcher

  (score
    [this term text]
    (let [get-sentences (:get-sentences this)
          tokenizer (:tokenize this)
          pos-tagger (:pos-tag this)
          score-words (get-scored-terms text term get-sentences tokenizer pos-tagger)]
      (score-text text score-words get-sentences tokenizer)))

  (rank
    [this term text] 
    (let [get-sentences (:get-sentences this)
          tokenizer (:tokenize this)
          pos-tagger (:pos-tag this)
          score-words (get-scored-terms text term get-sentences tokenizer pos-tagger)]
      (reverse (sort-by second (score-sentences text score-words get-sentences tokenizer))))))


(defn make-context-searcher
  "Generate a new Context Searcher using the given models. 3 models are
  required, a sentence detector model, a tokenizing model and a pos-tagging
  model."
  [smodel tmodel pmodel]
  (let [get-sentences (nlp/make-sentence-detector smodel)
        tokenizer (nlp/make-tokenizer tmodel)
        pos-tagger (nlp/make-pos-tagger pmodel)]
    (ContextSearcher. get-sentences tokenizer pos-tagger)))


; How to refer to the 'score' and 'rank' methods?

;user=> (use 'clomoios.contextsearcher)
;nil
;
;user=>  (def foo (make-context-searcher "models/EnglishSD.bin.gz" "models/EnglishTok.bin.gz" "models/tag.bin.gz"))
;#'user/foo
;
;user=> foo
;#:clomoios.contextsearcher.ContextSearcher{:get-sentences #<nlp$make_sentence_detector__248$sentenizer__249 opennlp.nlp$make_sentence_detector__248$sentenizer__249@766b62d1>, :tokenize #<nlp$make_tokenizer__252$tokenizer__253 opennlp.nlp$make_tokenizer__252$tokenizer__253@9c4ff2c>, :pos-tag #<nlp$make_pos_tagger__258$pos_tagger__259 opennlp.nlp$make_pos_tagger__258$pos_tagger__259@64552a2b>}
;
;user=> (score foo "a" "a")
;java.lang.Exception: Unable to resolve symbol: score in this context (NO_SOURCE_FILE:6)
;
;user=> (clomoios.contextsearcher/score foo "a" "a")
;java.lang.Exception: No such var: clomoios.contextsearcher/score (NO_SOURCE_FILE:7)
;
;user=> (clomoios.contextsearcher.ContextSearcher/score foo "a" "a")
;java.lang.IllegalArgumentException: No matching method: score (NO_SOURCE_FILE:8)