julia> using ConceptnetNumberbatch, Languages
       file_conceptnet = download_embeddings(url=CONCEPTNET_HDF5_LINK,
                                             localfile="./_conceptnet_/conceptnet.h5");
# [ Info: Download ConceptNetNumberbatch to ./_conceptnet_/conceptnet.h5...
#   % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
#                                  Dload  Upload   Total   Spent    Left  Speed
# 100  127M  100  127M    0     0  3646k      0  0:00:35  0:00:35 --:--:-- 4107k
# "./_conceptnet_/conceptnet.h5"

# Load embeddings
julia> conceptnet = load_embeddings(file_conceptnet, languages=:en)
# ConceptNet{Languages.English} (compressed): 1 language(s), 150875 embeddings

julia> conceptnet["apple"]  # Get embeddings for a single word
# 300-element Array{Int8,1}:
#   0
#   0
#   1
#  -4
# ...

julia> conceptnet[["apple", "pear", "cherry"]]  # Get embeddings for multiple words
# 300×3 Array{Int8,2}:
#   0   0   0
#   0   0   0
#   1   1   1
#  -4  -6  -7
# ...

# Load multiple languages
julia> conceptnet = load_embeddings(file_conceptnet, languages=[:en, :fr])
# ConceptNet{Language} (compressed): 2 language(s), 174184 embeddings

julia> conceptnet["apple"]  # fails, language must be specified
# ERROR: ...

julia> [conceptnet[:en, "apple"] conceptnet[:fr, "poire"]]
# 300×2 Array{Int8,2}:
#   0   -2
#   0   -2
#   1   -2
#  -4   -7
# ...

# Wildcard matching
julia> conceptnet[:en, "xxyyzish"]  # returns embedding for "#####ish"
# 300×1 Array{Int8,2}:
#   5
#  -1
#   0
#   1
# ...

# Useful functions
julia> length(conceptnet)  # total number of embeddings for all languages
# 174184

julia> size(conceptnet)  # embedding vector length, number of embeddings
# (300, 174184)

julia> "apple" in conceptnet  # found in the English embeddings
# true

julia> "poire" in conceptnet  # found in the French embeddings
# true

julia> # `keys` returns an iterator for all words
       for word in Iterators.take(keys(conceptnet),3)
           println(word)
       end
# définie
# invités
# couvents

julia> doc = "embed this document containing X_#-s231 which cannot be embedded"
       edoc, idxs_missed = embed_document(conceptnet, doc, language=Languages.English(), keep_size=false)
       missed_words = tokenize_for_conceptnet(doc)[idx_missed]
       println("Missed word: $missed_word")
       edoc
# Missed word: SubString{String}["X_#-s231"]
# 300×8 Array{Int8,2}:
#   0   0   0   0   0   1   0   0
#  -1  -2  -1  -1  -3  -2  -3   0
#   1   5   0   4   6   6   6   2
# ...