bigram.sql

SELECT
  string_agg(token, ' ') as ngram,
  count(*) as count,
  review_id,
  MIN(start) AS start, MAX("end") AS end,
  CASE
  WHEN token = 'guy' THEN
    index + 1
  WHEN token = 'stand' THEN
    unnest(ARRAY[index, index+1])
  ELSE
    index
  END AS _index
FROM
  index
WHERE
  city = 'Tempe' AND
  token IN ('guy', 'stand', 'behind') 
GROUP BY _index, review_id
HAVING string_agg(token, ' ') in ('guy stand', 'stand behind')
ORDER BY count