![]() ('sycophant', 'flatterer', 0.343303918838501)Īny thoughts on this? What am I doing incorrectly? My tokenization is standard nltk word_tokenize on a string of only words. Visual inspection of the list of tuples indicates a very low cosine relation - while my colleague using the same approach achieved 0.650, which seems complementary to my output. Similarity_scores.sort(key=lambda x: x, reverse=True) likeness resemblance reproduction suggest new resemblance n. Similarity_scores.append((array_word1, array_word2, If array_word1 in embeddings_dict and array_word2 in embeddings_dict: #find matches for both words Similarity_scores = # declare a list to hold similarity score resultsįor array_word1 in array1:# iterate over words in array1 in order to compareįor array_word2 in array2: #iterate over words in array2 in order to compare Note that I am comparing two arrays, but am concerned about the output for all synonyms: def compare_embeddings(array1, array2): With open("glove.6b/glove.6B.300d.txt", 'r', encoding="utf-8") as f: # load the file into memory Here is where I create my embeddings dictionary, and notice that it is a fairly high dimension vector space (300d): def embedding_dictionary():Įmbeddings_dict = # create a dictionary Wondering if there might be something I've done incorrectly? The words are: ('sycophant', 'flatterer') ![]() I receive a value closer to zero than one, which seems odd, considering they are synonymous. Try using a synonym dictionary if you're lost for words to try! You might come across something even closer to the goal.Working with GloVe embeddings, I am receiving odd output (and consistently different from a colleague) for words that are definitely synonyms.It feels tricky at first, but after completing a puzzle or two, you'll get the hang of how to assess words and their meanings in relation to the goal word. Quordle is a five-letter word guessing game similar to Wordle, except each guess applies letters to four words at the same time. ![]() Words are case-sensitive - don't use capital letters unless you are guessing a proper noun.Semantle has a list of the 1,000 closest words to the goal word - when you type one of those words in, you'll know how many words away from the goal word it is by the fourth column along.The higher the Similarity score of a word, the closer it is to the target word semantically.Good starting words include apple, work, building, sport, fight, and small. Most like Similar Least like Cannabis is found in 9 of listing names Medical is closely linked to the cannabis area Beach vibes Words like cove, ocean and caribbean Spanish words like bodega and. Try a range of different words at the beginning to get a feel for what the word could be. On this page youll find 51 synonyms, antonyms, and words related to similarities, such as: comparison, resemblance, relationship, closeness, coincidence. ![]() For example, if the target word is 'coin,' words like 'pay' or 'historical' would likely nudge you closer to the goal.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |