I'm new to machine learning and I'm trying to come up with a model that will complete all second words in phrases. I couldn't find solution to this exact problem although there are lots of tutorials on generating text with RNN.
So, consider you have the 2 following files:
1) a word dictionary for training
Say we have a table with 2 columns of word pairs: 'complete' and 'sample' such that the first column includes different word pairs ("Hello dear", "my name", "What time", "He goes", etc.) and the second one includes first words and only a part (> 2 letters) of second words ("Hello de", "my nam", "What ti", "He goe", etc.).
2) a table for testing
It's a table that consists of only 'sample' column.
The aim is to add 'complete' column to the second table with complete pairs of words.
I came up with the only way to do this:
compute the frequences of all first words (P(w1))
compute the frequences of all complete second words (P(w2))
compute the frequences of all first words given complete second words (P(w1|w2))
predict complete second words using Bayes rule:
w2 = argmax_{w2} ( P(w2|w1)) = argmax_{w2} (P(w1|w2) * P(w2))
for each w1 in the test table w2 is the most probable w2 or the most frequent w2 (if w1 is not in the dict).
The problem is this algorithm doesn't work sufficiently well. How can I somehow optimise the probabilities (maybe gradient descent might be helpful?)? Is there any other way to address this task?
Related
I am confused about the meaning of paragraph matrix (D) in the structure of the PV-DBOW model (Doc2Vec).
Is the paragraph matrix the result of one-hot encoding of n input paragraph IDs?
Or is the paragraph matrix a randomly initialized weight to generate the shape(nxp), where n is the number of paragraph ID inputs and p is the vector dimension?
I've never found the original paper's diagrams very clear or helpful. (And, no one has been able to reproduce their claimed results in the 'concatenation' mode.)
But I can say, from familiarity with code implementations:
At the outset, every paragraph-ID gets a randomly-initialized vector. Thus, there is a matrix in the model with all of these vectors, of shape number_of_paragraph_ids x number_of_dimensions
For backprop-training in PV-DBOW, each individual paragraph-vector (one row from the above matrix) is adjusted to better predict the matching-paragraph's individual constituent words.
While figuratively, that's sort of a 1-hot selection of a single paragraph choice, in the code it's just a lookup of the single correct row using a paragraph-ID key.
I am trying to build a classifier to classify some files into 150 categories based on the name of those files. Here are some examples of file names in my dataset (~700k files):
104932489 - urgent - contract validation for xyz limited.msg
treatment - an I l - contract n°4934283 received by partner.pdf
- invoice_8843238_1_europe services_business 8592342sid paris.xls
140159498736656.txt
140159498736843.txt
fsk_000000001090296_sdiselacrusefeyre_2000912.xls
fsk_000000001091293_lidlsnd1753mdeas_2009316.xls
You can see that the filenames can really be anything, but that however there is always some pattern that is respected for the same categories. It can be in the numbers (that are sometimes close), in the special characters (spaces, -, °), sometimes the length, etc.
Extracting all those patterns one by one will take ages because I have approximately 700k documents. Also, I am not interested in 100% accuracy, 70% can be good enough.
The real problem is that I don't know how to encode this data. I have tried many methods:
Tokenizing character by character and feeding them to an LSTM model with an embedding layer. However, I wasn't able to implement it and got dimension errors.
Adapting Word2Vec to convert the characters into vectors. However, this automatically drops all punctuation and space characters, also, I lose the numeric data. Another problem is that it creates more useless dimensions: if the size is 20, I will have my data in 20 dimensions but if I look closely, there are always the same 150 vectors in those 20 dimensions so it's really useless. I could use a 2 dimensions size but still, I need the numeric data and the special characters.
Generating n-grams from each path, in the range 1-4, then using a CountVectorizer to compute the frequencies. I checked and special characters were not dropped but it gave me like 400,000 features! I am running a dimensionality reduction using UMAP (n_components=5, metric='hellinger') but the reduction runs for 2 hours and then the kernel crashes.
Any ideas?
I am currently also working on a character level lstm. And it works exactly the same like when you would use words. You need a vocabulary, for example a - z and then you just take the index of the letter as its integer representation. For example:
"bad" -> "b", "a", "d" -> [1, 0, 3]
Now you could create an embedding lookup table (for example using pytorchs nn.Embedding function). You just have to create a random vector for every index of your vocab. For example:
"a" -> 0 > [-0.93, 0.024, -.0.73, ..., -0.12]
You said that you tried this but encountered dimension errors? Maybe show us the code!
Or you could create non-random embedding using word2vec using the Gensim libary:
from gensim.models import Word2Vec
# 'total_words' is a list containing every word of your dataset split into its characters
total_words = [...]
model = Word2Vec(total_words , min_count=1, size=32)
model.save(save_model_file)
# lets test it for the character 'a'
embedder = Word2Vec.load(save_model_file)
v = embedder["a"]
# v now will be a the embedding vector of a with size 32x1
I hope I could make clear how to create embeddings for characters.
You can treat characters in single-word-classification the exact same way you would treat words in sentence-classification.
I have two text datasets. Each dataset consists of multiple sequences and each sequence can have more than one sentence.
How do I measure if both datasets are from same distribution?
The purpose is to verify transfer learning from one distribution to another only if the difference between the distributions is statistically significant.
I am panning to use chi-square test but not sure if it will help for text data considering the high degrees of freedom.
update:
Example:
Supppose I want to train a sentiment classification model. I train a model on IMDb dataset and evaluate on IMDb and Yelp datasets. I found that my model trained on IMDb still does well on Yelp. But the question is how different these datasets are?
Train Dataset : https://www.kaggle.com/columbine/imdb-dataset-sentiment-analysis-in-csv-format?select=Train.csv
Eval 1: https://www.kaggle.com/columbine/imdb-dataset-sentiment-analysis-in-csv-format?select=Valid.csv
Eval 2: https://www.kaggle.com/omkarsabnis/sentiment-analysis-on-the-yelp-reviews-dataset
Now,
How different are train and eval 1?
How different are train and eval 2?
Is the dissimilarity between train and eval 2 by chance ? What is the statistical significance and p value?
The question "are text A and text B coming from the same distribution?" is somehow poorly defined. For example, these two questions (1,2) can be viewed as generated from the same distribution (distribution of all questions on StackExchange) or from different distributions (distribution of two different subdomains of StackExchange). So it's not clear what is the property that you want to test.
Anyway, you can come up with any test statistic of your choice, approximate its distribution in case of "single source" by simulation, and calculate the p-value of your test.
As a toy example, let's take two small corpora: two random articles from English Wikipedia. I'll do it in Python
import requests
from bs4 import BeautifulSoup
urls = [
'https://en.wikipedia.org/wiki/Nanjing_(Liao_dynasty)',
'https://en.wikipedia.org/wiki/United_States_Passport_Card'
]
texts = [BeautifulSoup(requests.get(u).text).find('div', {'class': 'mw-parser-output'}).text for u in urls]
Now I use a primitive tokenizer to count individual words in texts, and use root mean squared difference in word relative frequencies as my test statistic. You can use any other statistic, as long as you calculate it consistently.
import re
from collections import Counter
from copy import deepcopy
TOKEN = re.compile(r'([^\W\d]+|\d+|[^\w\s])')
counters = [Counter(re.findall(TOKEN, t)) for t in texts]
print([sum(c.values()) for c in counters])
# [5068, 4053]: texts are of approximately the same size
def word_freq_rmse(c1, c2):
result = 0
vocab = set(c1.keys()).union(set(c2.keys()))
n1, n2 = sum(c1.values()), sum(c2.values())
n = len(vocab)
for word in vocab:
result += (c1[word]/n1 - c2[word]/n2)**2 / n
return result**0.5
print(word_freq_rmse(*counters))
# rmse is 0.001178, but is this a small or large difference?
I get a value of 0.001178, but I don't know whether it's a large difference. So I need to simulate the distribution of this test statistic under the null hypothesis: when both texts are from the same distribution. To simulate it, I merge two texts into one, and then split them randomly, and calculate my statistic when comparing these two random parts.
import random
tokens = [tok for t in texts for tok in re.findall(TOKEN, t)]
split = sum(counters[0].values())
distribution = []
for i in range(1000):
random.shuffle(tokens)
c1 = Counter(tokens[:split])
c2 = Counter(tokens[split:])
distribution.append(word_freq_rmse(c1, c2))
Now I can see how unusual is the value of my observed test statistic under the null hypothesis:
observed = word_freq_rmse(*counters)
p_value = sum(x >= observed for x in distribution) / len(distribution)
print(p_value) # it is 0.0
print(observed, max(distribution), sum(distribution) / len(distribution)) # 0.0011 0.0006 0.0004
We see that when texts are from the same distribution, my test statistic is on average 0.0004 and almost never exceeds 0.0006, so the value of 0.0011 is very unusual, and the null hypothesis that two my texts originate from the same distribution should be rejected.
I wrote an article which is similar to your problem but not exactly the same.
https://towardsdatascience.com/a-new-way-to-bow-analysis-feature-engineering-part1-e012eba90ef
The problem that I was trying to solve is to check if a word has different (significant) distributions across categories or labels.
There are a few similarities between your problem and the one I had mentioned above.
You want to compare two sources of datasets, which can be taken as two different categories
Also, to compare the data sources, you will have to compare the words as sentences can't be directly compared
So, my proposed solution to this will be as:
Create words features across the two datasets using count-vectorizer and get top X words from each
Let's say you have total distinct words as N, now initialize count=0 and start to compare the distribution for each word and if the differences are significant increment the counter. Also, there could be cases where a word only exists in one of the datasets and that is a good new, by that I mean it shows that it is a distinguishing feature, so, for this also increment the count
Let's say the total count is n. Now, the lower is the n/N ratio, similar two texts are and vice-a-versa
Also, to verify this methodology - Split the data from a single source into two (random sampling) and run the above analysis, if the n/N ratio is closer to 0 which indicates that the two data sources are similar which also is the case.
Please let me know if this approach worked or not, also if you think there are any flaws in this, I would love to think and try evolving it.
This Article describes the LambdaRank algorithm for information retrieval. In formula 8 page 6, the authors propose to multiply the gradient (lambda) by a term called |∆NDCG|.
I do understand that this term is the difference of two NDCGs when swapping two elements in the list:
the size of the change in NDCG (|∆NDCG|) given by swapping the rank positions of U1 and U2
(while leaving the rank positions of all other urls unchanged)
However, I do not understand which ordered list is considered when swapping U1 and U2. Is it the list ordered by the predictions from the model at the current iteration ? Or is it the list ordered by the ground-truth labels of the documents ? Or maybe, the list of the predictions from the model at the previous iteration as suggested by Tie-Yan Liu in his book Learning to Rank for Information Retrieval ?
Short answer: It's the list ordered by the predictions from the model at the current iteration.
Let's see why it makes sense.
At each training iteration, we perform the following steps (these steps are standard for all Machine Learning algorithms, whether it's classification or regression or ranking tasks):
Calculate scores s[i] = f(x[i]) returned by our model for each document i.
Calculate the gradients of model's weights ∂C/∂w, back-propagated from RankNet's cost C. This gradient is the sum of all pairwise gradients ∂C[i, j]/∂w, calculated for each document's pair (i, j).
Perform a gradient ascent step (i.e. w := w + u * ∂C/∂w where u is step size).
In "Speeding up RankNet" paragraph, the notion λ[i] was introduced as contributions of each document's computed scores (using the model weights at current iteration) to the overall gradient ∂C/∂w (at current iteration). If we order our list of documents by the scores from the model at current iteration, each λ[i] can be thought of as "arrows" attached to each document, the sign of which tells us to which direction, up or down, that document should be moved to increase NDCG. Again, NCDG is computed from the order, predicted by our model.
Now, the problem is that the lambdas λ[i, j] for the pair (i, j) contributes equally to the overall gradient. That means the rankings of documents below, let’s say, 100th position is given equal improtance to the rankings of the top documents. This is not what we want: we should prioritize having relevant documents at the very top much more than having correct ranking below 100th position.
That's why we multiply each of those "arrows" by |∆NDCG| to emphasise on top ranking more than the ranking at the bottom of our list.
I am currently working on a machine learning project, and am in the process of building the dataset. The dataset will be comprised of a number of different textual features, of varying length from 1 sentence to around 50 sentences(including punctuation). What is the best way to store this data to then pre-process and use for machine learning using python?
In most cases, you can use a method called Bag of Word, however, in some cases when you are performing more complicated task like similarity extraction or want to make comparison between sentences, you should use Word2Vec
Bag of Word
You may use the classical Bag-Of-Word representation, in which you encode each sample into a long vector indicating the count of all the words from all samples. For example, if you have two samples:
"I like apple, and she likes apple and banana.",
"I love dogs but Sara prefer cats.".
Then all the possible words are(order doesn't matter here):
I she Sara like likes love prefer and but apple banana dogs cats , .
Then the two samples will be encoded to
First: 1 1 0 1 1 0 0 2 0 2 1 0 0 1 1
Second: 1 0 1 0 0 1 1 0 1 0 0 1 1 0 1
If you are using sklearn, the task would be as simple as:
from sklearn.feature_extraction.text import CountVectorizer
vectorizer = CountVectorizer()
corpus = [
'This is the first document.',
'This is the second second document.',
'And the third one.',
'Is this the first document?',
]
X = vectorizer.fit_transform(corpus)
# Now you can feed X into any other machine learning algorithms.
Word2Vec
Word2Vec is a more complicated method, which attempts to find the relationship between words by training a embedding neural network underneath. An embedding, in plain english, can be thought of the mathematical representation of a word, in the context of all the samples provided. The core idea is that words are similar if their contexts are similar.
The result of Word2Vec are the vector representation(embeddings) of all the words shown in all the samples. The amazing thing is that we can perform algorithmic operations on the vector. A cool example is: Queen - Woman + Man = King reference here
To use Word2Vec, we can use a package called gensim, here is a basic setup:
model = Word2Vec(sentences, size=100, window=5, min_count=5, workers=4)
model.most_similar(positive=['woman', 'king'], negative=['man'])
>>> [('queen', 0.50882536), ...]
Here sentences is your data, size is the dimension of the embeddings, the larger size is, the more space is used to represent a word, and there is always overfitting we should think about. window is the size of the context we are cared about, it is the number of words before the target word we are looking at when we are predicting the target from its context, when training.
One common way is to create your dictionary(all the posible words) and then encode every of your examples in function of this dictonary, for example(this is a very small and limited dictionary just for example) you could have a dictionary : hello ,world, from, python . Every word will be associated to a position, and in every of your examples you define a vector with 0 for inexistence and 1 for existence, for example for the example "hello python" you would encode it as: 1,0,0,1