Reproducibility / Working examples

[Querela]

Hi, this seems like a very interesting project. Unfortunately, I can't seem to get any results. I'm a bit unsure why and how to solve this.

• The screenshot in the README img_1.png does not really reflect the current API.
• The screenshot img_4.png and the included example example.py also do not produce any output.
• Trying to install the package, it seems like spacy and pandas are missing. I suggest, including them into the setup.py. E.g. like this.
• (Maybe also add a comment about the spacy model installation, e.g. python -m spacy download de_core_news_lg?)

I tried with both patterns_small.jsonl and patterns_large.jsonl. And a few different inputs...

some tests

I just extended the `example.py` file: ```python # ... for sent in doc.sents: print(f"In: {sent}") widths = [max(6, len(word.text)) for word in sent] fmt = lambda x: f"{x[0]:{x[1]}}" fmtit = lambda x: map(fmt, zip(x, widths)) print("Text:", " ".join(fmtit(map(str, (word.text for word in sent))))) print("pos_:", " ".join(fmtit(map(str, (word.pos_ for word in sent))))) print("tag_:", " ".join(fmtit(map(str, (word.tag_ for word in sent))))) print("dep_:", " ".join(fmtit(map(str, (word.dep_ for word in sent))))) print("part:", " ".join(fmtit(map(str, (word._.part for word in sent))))) print("posi:", " ".join(fmtit(map(str, (word._.posi for word in sent))))) # ... ``` ``` In: Nürnberg ist eine Stadt in Deutschland. Text: Nürnberg ist eine Stadt in Deutschland . pos_: PROPN AUX DET NOUN ADP PROPN PUNCT tag_: NE VAFIN ART NN APPR NE $. dep_: sb ROOT nk pd mnr nk punct part: subj pred obj obj pred subj None posi: 1 0 0 2 0 0 None # no triples ``` ``` In: Rosa Kempf schloss 1911 ihr Studium mit der Promotion ab Text: Rosa Kempf schloss 1911 ihr Studium mit der Promotion ab pos_: PROPN PROPN VERB NUM DET NOUN ADP DET NOUN ADP tag_: NE NE VVFIN CARD PPOSAT NN APPR ART NN PTKVZ dep_: pnc sb ROOT mo nk oa mnr nk nk svp part: subj subj pred None None obj None None None None posi: 0 6 0 None None 1 None None None None # no triples ```

[ChrisDelClea]

Hi @Querela ,

thanks for your questions and recommendations .

• You mentioned that pattern_builder shown in img_1.png does not reflect the current API. What exactly does not fit in your opinion? I tried to find any error on my end, but in my demo it still works like it is described .
• Regarding the second point, however, I indeed made some small changes to API, sorry for that. I updated the readme.
• I also updated the setup.py as you recommended and included the python installation command for the spacy model I used in the setup section of the readme.

You might want to test it again. If you have any other issues let me know!

Best regards Chris