Evidence

Local setup

Create single data directory for all models / input data

mkdir data

Obtain TargetSize150.zip, unpack and move into data directory

unzip -qq TargetSize150.zip && mv TargetSize150/* data && rmdir TargetSize150

Obtain doc2vec.csv, then store in data directory

mv doc2vec.csv data

Run

After local deploy (requires ./data setup as per "Local setup" instructions)

docker-compose up --build

Add a user:

curl -XPOST http://localhost:3000/users -d '${username}'

Elastic search example queries

List of snippets:

http://localhost:8080/es/snippets/_search

Single snippet:

http://localhost:8080/es/snippets/snippet/$id

More like this:

curl -XGET -H 'Content-Type: application/json' \
    http://localhost:8080/es/snippets/_search -d '{
    "query": {
        "more_like_this": {
            "fields": ["text", "lemma"],
            "boost_terms": 1,
            "max_query_terms": 150,
            "min_doc_freq": 1,
            "min_term_freq": 1,
            "like": [{
                "_index": "snippets",
                "_type": "snippet",
                "_id": "'${id}'"
            }]
        }
    }
}'

Doc2Vec example queries

More like this:

curl http://localhost:8080/doc2vec/$id

Paging in doc2vec is done using request parameters from (default 0) and size (default 10)

curl http://localhost:8080/doc2vec/$id?from=3&size=8

Legal matters

Copyright 2016-2019 Koninklijke Nederlandse Academie van Wetenschappen

Distributed under the terms of the GNU General Public License, version 3. See the file LICENSE for details.