Collocations

collocations(text="", path=null, window=5, measure="logdice", min_freq=2, node=null, lemmatize=true, top_n=20)

Collocations - pairs of words that occur together more often than by chance: set phrases, terms, the collocates of a word (node). A pair is ordered: the right word stands at most window words after the left one. With lemmatize lemmas are compared, without disambiguation: «точка зрения» and «точки зрения» are one pair. The definitions and formulas of the measures are on the ruTS page Collocations.

The pair frequency freq_pair is the number of position pairs where the right word stands within the window after the left one. A word repeated within the window gives several pairs, so freq_pair can exceed the frequency of a word, and such a pair is a repetition rather than a set phrase.

Errors - those common to the tools, a text without words and a node that is not one word.

CallResult

{
  "name": "collocations",
  "arguments": {
    "text": "Кот сидел на окне и смотрел на птиц. Птицы улетели, и кот уснул на окне. Завтра кот снова будет сидеть на окне и смотреть на птиц.",
    "top_n": 3
  }
}

{
  "n_words": 26,
  "measure": "logdice: logDice (Rychlý 2008): чем выше, тем устойчивее пара, не зависит от размера текста; пара, которая всегда стоит рядом, получает 11.7, наибольшее значение 14 - у пары, которая встречается на каждом расстоянии в окне, ниже нуля - слабая связь",
  "collocations": [
    {"left": "на", "right": "птица", "freq_left": 5, "freq_right": 3, "freq_pair": 5, "score": 12.0},
    {
      "left": "смотреть",
      "right": "птица",
      "freq_left": 2,
      "freq_right": 3,
      "freq_pair": 3,
      "score": 11.94
    },
    {
      "left": "сидеть",
      "right": "на",
      "freq_left": 2,
      "freq_right": 5,
      "freq_pair": 4,
      "score": 11.87
    }
  ],
  "warnings": []
}

Parameters

Parameter	Type	Default	Description
text	string	""	Text
path	string	null	Absolute path to a file with the text instead of text
window	integer	5	Greatest distance between the words of a pair, from 1 to 10; 1 - adjacent words
measure	string	"logdice"	Association measure: logdice, mi, mi3, t_score, dice, log_likelihood, npmi, min_sensitivity; logdice does not depend on the text size, mi overrates rare pairs, mi3 and t_score favor frequent ones
min_freq	integer	2	Least pair frequency freq_pair
node	string	null	Word whose collocates are needed; not given - all pairs
lemmatize	boolean	true	Compare lemmas
top_n	integer	20	Number of pairs in the response, up to 200

Result

Field	Type	Description
n_words	integer	Number of words in the text
measure	string	The measure and how to read its values
collocations	array[object]	Pairs by descending measure: the left word left, the right one right, their frequencies freq_left and freq_right, the pair frequency freq_pair and the value of the measure score
warnings	array[string]	The text is not in Russian, the node word is not in the text, there are no pairs with such a frequency