Coverage report: /home/ellis/comp/core/lib/nlp/section.lisp

Kind	Covered	All	%
expression	43	46	93.5
branch	1	2	50.0

Key

Not instrumented

Conditionalized out

Executed

Not executed

Both branches taken

One branch taken

Neither branch taken

1

(in-package :nlp/section)

2

3

(defun extract-sections (text &key (epsilon 0.5))

4

"Extract the sections from a string of text. Epsilon refers to the

5

distance between two points for them to be considered related."

6

(labels ((average-distance (point points)

7

(/ (reduce #'+ points

8

:key (lambda (i) (distance (vector-data i)

9

                                                    (vector-data point))))

10

(length points))))

11

(let ((collection (make-instance 'document-collection)))

12

(loop for sentence in (sentence-tokenize text)

13

do (add-document collection

14

(make-instance 'document-cluster

15

:string-contents sentence)))

16

(tf-vectorize-documents collection)

17

(loop for document in (documents collection)

18

with cluster-index = 0

19

for cluster = (get-cluster cluster-index (documents collection))

20

do (if (and cluster (>= epsilon (average-distance document cluster)))

21

(setf (cluster document) cluster-index)

22

(setf (cluster document) (incf cluster-index))))

23

collection)))