Analysis of FSF Philosophy

There are 218 English-language articles on philosophy on gnu.org. A sentiment analysis of all of the sentences from this corpus suggests that the GNU philosophy is 4x more negative about commerce than about community. Here are the top 30 pages on the topics of users, community, and commerce:

i	Page	Pagerank	All Terms	User	Community	Commerce
1	free-sw	56	48	25	4	19
2	free-software-even-more-important	30	50	33	2	15
3	words-to-avoid	26	110	46	9	55
4	open-source-misses-the-point	25	51	33	8	10
5	who-does-that-server-really-serve	20	115	30	2	83
6	categories	18	34	11	1	22
7	not-ipr	14	5	2	0	3
8	why-call-it-the-swindle	13	13	9	0	4
9	javascript-trap	9	21	18	2	1
10	surveillance-vs-democracy	8	51	21	0	30
11	copyright-and-globalization	3	90	9	2	79
12	copyright-versus-community	3	84	22	23	39
13	rms-nyu-2001-transcript	2	187	45	42	100
14	copyright-versus-community-2000	2	65	5	15	45
15	pragmatic	2	21	5	12	4
16	proprietary-surveillance	1	363	274	0	89
17	stallman-mec-india	1	123	21	15	87
18	free-digital-society	1	113	40	1	72
19	google-engineering-talk	1	108	62	11	35
20	malware-microsoft	1	99	85	0	14
21	nit-india	1	95	43	7	45
22	danger-of-software-patents	1	94	13	2	79
23	malware-apple	1	89	76	0	13
24	ough-interview	1	87	29	16	42
25	rms-hack	1	72	2	32	38
26	moglen-harvard-speech-2004	1	54	6	2	46
27	rms-interview-edinburgh	1	46	16	10	20
28	android-and-users-freedom	1	44	43	0	1
29	digital-inclusion-in-freedom	0	69	42	1	26
30	free-software-for-freedom	0	54	17	12	25

How to Run

Depends on wget (thank you GNU! :) and Python 3.

python3 -m venv .env --prompt='fsf'
source .env/bin/activate
pip install --upgrade --requirement requirements.txt

wget --mirror --no-parent https://www.gnu.org/philosophy/
find www.gnu.org/philosophy -name \*.html \
  | sed -e 's/^www.gnu.org\/philosophy\///' \
  > html.txt
./filter-en.py > html-en.txt
./rank.py > ranked.txt
./find-sentences.py > sentences.txt
./prep-csv.py > sentences.csv

# https://huggingface.co/blog/sentiment-analysis-python
time ./label-sentiment.py # modifies sentences.csv

./report.py

Dropped 231 sentences out of 2129 for low confidence, 1898 remaining.

           neg /  pos
commerce  1288 /  284 = 5
community  180 /  146 = 1
                     /= 4

The GNU philosophy is 4x more negative about commerce than about community.