Word | Frequency | Number of right neighbors | Number of left neighbors | Ratio |
---|---|---|---|---|
But | 733 | 35 | 2 | 17.5000 |
On | 218 | 17 | 1 | 17.0000 |
The | 4783 | 346 | 22 | 15.7273 |
And | 416 | 27 | 2 | 13.5000 |
We | 703 | 47 | 4 | 11.7500 |
He | 920 | 44 | 4 | 11.0000 |
That's | 105 | 11 | 1 | 11.0000 |
For | 221 | 21 | 2 | 10.5000 |
They | 451 | 31 | 3 | 10.3333 |
It’s | 111 | 10 | 1 | 10.0000 |
York | 180 | 9 | 1 | 9.0000 |
It's | 350 | 26 | 3 | 8.6667 |
In | 883 | 58 | 7 | 8.2857 |
States | 188 | 8 | 1 | 8.0000 |
She | 315 | 15 | 2 | 7.5000 |
While | 138 | 7 | 1 | 7.0000 |
am | 111 | 7 | 1 | 7.0000 |
candidates | 66 | 7 | 1 | 7.0000 |
It | 806 | 41 | 6 | 6.8333 |
So | 191 | 12 | 2 | 6.0000 |
Word | Frequency | Number of right neighbors | Number of left neighbors | Ratio |
---|---|---|---|---|
expected | 181 | 1 | 9 | 0.1111 |
kind | 118 | 1 | 9 | 0.1111 |
part | 359 | 3 | 27 | 0.1111 |
wants | 99 | 1 | 8 | 0.1250 |
allowed | 67 | 1 | 8 | 0.1250 |
29 | 52 | 1 | 7 | 0.1429 |
Department | 152 | 2 | 13 | 0.1538 |
years ago | 86 | 2 | 12 | 0.1667 |
pretty | 60 | 1 | 6 | 0.1667 |
responsible | 66 | 1 | 6 | 0.1667 |
track | 61 | 1 | 6 | 0.1667 |
strike | 47 | 1 | 6 | 0.1667 |
be an | 31 | 1 | 5 | 0.2000 |
amount | 86 | 2 | 10 | 0.2000 |
Taliban | 48 | 1 | 5 | 0.2000 |
limited | 39 | 1 | 5 | 0.2000 |
NBC | 49 | 1 | 5 | 0.2000 |
wasn't | 80 | 1 | 5 | 0.2000 |
according | 295 | 1 | 5 | 0.2000 |
expensive | 35 | 1 | 5 | 0.2000 |
In this subsection, we compute the ratio of the number of right neighbors and the number of left neighbors. Again, we look for words with extreme ratios:
Data for first table:
select word,w.freq,aa.cnt, bb.cnt,aa.cnt/bb.cnt as r from words w, (select w1_id,count(c.w2_id) as cnt from co_n c where w1_id>100 group by w1_id) aa, (select w2_id,count(c.w1_id) as cnt from co_n c where w2_id>100 group by w2_id) bb where w_id=aa.w1_id and aa.w1_id=bb.w2_id order by r desc limit 20;
Diagram data:
select aa.cnt, bb.cnt from (select w1_id,count(c.w2_id) as cnt from co_n c where w1_id>100 group by w1_id) aa, (select w2_id,count(c.w1_id) as cnt from co_n c where w2_id>100 group by w2_id) bb where aa.w1_id=bb.w2_id;
5.1.7.1 Number of NN co-occurrences vs. Frequency I
5.1.7.2 Number of NN co-occurrences vs. Frequency II