Questions tagged [nlp]

Natural language processing (NLP) is a field of computer science, artificial intelligence, and linguistics concerned with the interactions between computers and human (natural) languages. As such, NLP is related to the area of human–computer interaction. Many challenges in NLP involve natural language understanding, that is, enabling computers to derive meaning from human or natural language input, and others involve natural language generation.

Filter by
Sorted by
Tagged with
0 votes
1 answer
67 views

What are popular deep learning models for tabular data of texts?

I have a tabular dataset where every column is of type "text" (i.e. not categorical variable and essentially it can be anything). Let's suppose that the task is classification What are some ...
user avatar
  • 141
0 votes
0 answers
24 views

One class classifier for fraud call detection ( in Hindi language) using BERT

I have created a dataset of text files that are nothing but transcripts of fraud call recordings. I want to implement one class classifier as I have only fraud call audio and transcripts but not a ...
user avatar
0 votes
1 answer
37 views

Get the keywords from positive and negative reviews

I have trained a classifier algorithm on a sentiment analysis model which classifies the reviews scraped off Amazon as Positive or Negative. Now for each class, I want to get the keywords from the ...
user avatar
  • 1,231
0 votes
1 answer
24 views

Making my own stop-words list from a certain community, is tf-idf good enough?

So I have some tweets from my country and I want to make a my own stop-words list. Is tf-idf good enough? Are there any statistical methods that would be better?
user avatar
  • 45
0 votes
0 answers
15 views

Algorithms for classification of very short text

I am to create a classification model for texts that typically have 3 to 4 words in them. I thought of using BERT and XLNet but not sure if they are the right choice for texts that short. Are there ...
user avatar
0 votes
1 answer
29 views

Document Content

I have a set of .pdf/.docx documents with content. I need to search for the most suitable document according to a particular sentence. For instance: ...
user avatar
0 votes
0 answers
13 views

How to prepare data for the encoder part of the model in chatbot using Tensorflow?

I hope you are doing great. I am trying to build a chatbot using Tensorflow. The dataset I am using is cornell movie dialogs. I have a very basic doubt. I am almost done with the preprocessing part of ...
user avatar
1 vote
1 answer
124 views

Why not rule-based semantic role labelling?

I have recently found some interest in automatic semantic role labelling. Most introductory texts (e.g. Jurafsky and Martin, 2008) present approaches based on supervised machine learning, often using ...
user avatar
2 votes
1 answer
50 views

How to feed a Knowledge Base into Language Models?

I’m a CS undergrad trying to make my way into NLP Research. For some time, I have been wanting to incorporate "everyday commonsense reasoning" within the existing state-of-the-art Language ...
user avatar
0 votes
0 answers
7 views

Given a job viewed or applied

I am very new to analyzing text data and extracting information out of it. I need some suggestion help from the communnity. The dataset is from the job website where 'id' denotes the job id, 'abstract'...
user avatar
  • 105
1 vote
1 answer
64 views

Clustering text data based on sentiment?

I am scraping reviews off Amazon with the intent to perform sentiment analysis to classify them into positve, negative and neutral. Now the data I would get would be text and unlabeled. My approach to ...
user avatar
  • 1,231
0 votes
1 answer
18 views

How to multi label text Classification using Deep learning

I am new to the multi-label text classification using Deep learning, I have Data like this: ...
user avatar
0 votes
0 answers
15 views

If Bert can handle only 512 inputs. Why you can provide such long contexts in QA Pipeline?

For example, I use Pipeline from Huggingface Transformers to use a QA model card like this. ...
user avatar
0 votes
0 answers
15 views

Group related items by their description and tag each group. [Pen, Eraser] : Stationary

So have a list of data similar to the table below. It will be captured by a chatbot so I expect natural language but in the form of a structured command: Add {Qty} {item description} to {location} ID ...
user avatar
  • 101
0 votes
0 answers
11 views

Use case: How to find sentiment expressed towards the "phrase" in the sentence?

Suppose I have a data like this, (separated by |): ...
user avatar
0 votes
0 answers
16 views

Calculating customer satisfaction based on call transcripts

I have a massive dataset of call center transcripts and I want to rate them on a scale of 1-10 based on customer satisfaction. What’s the best way to go about that?
user avatar
0 votes
1 answer
56 views

extracting data from unstructured pdfs

I have about 200,000 PDFs made up of 20 different designs. i.e In an organization, different (20) departments issue monthly award submission requirements. Each department has its own document format. ...
user avatar
  • 103
2 votes
2 answers
94 views

How to improve language model ex: BERT on unseen text in training?

I am using pre-trained language model for binary classification. I fine-tune the model by training on data my downstream task. The results are good almost 98% F-measure. However, when I remove a ...
user avatar
  • 123
0 votes
0 answers
15 views

How to arrange ground-truth for anchor box representation in object detection

I am working on CharGrid and BERTGrid papers and have questions about bounding box regression decoder part. In the CharGrid paper, it states that there are two outputs from this branch: one with ...
user avatar
1 vote
1 answer
21 views

Treating Word Embeddings as Multivariate Gaussian Random Variables

I want to specify some probabilistic clustering model (such as a mixture model or lda) over words, and instead of using the traditional method of representing words as an indicator vector , I want to ...
user avatar
  • 23
0 votes
0 answers
30 views

How Flair (NER) works?

I have found multiples papers (or websites) about flair. All those papers describes how to use flair for NER. I didn't found any paper or (websites) that describe flair architecture and how it works. ...
user avatar
  • 267
0 votes
0 answers
4 views

Size Matrix features after applying 6 1D Kernels on one-hot encoded vectors

Suppose we are building the following model to build a neural network over one-hot encoded vectors of characters: For a given dataset, it’s not reasonable to read the whole text! So, we take some ...
user avatar
  • 179
0 votes
1 answer
22 views

How to improve document classification between two similar documents

I have a document classification problem where I need to classify whether a certain document is about real estate or not. I get a URL of a webpage from which I extract all the text and then using my ...
user avatar
1 vote
0 answers
206 views

Why do we use squeeze(1) at this model definition with PyTorch?

PyTorch noobie here. I'm following an online tutorial and there's a simple model definition as follows: ...
user avatar
  • 438
1 vote
1 answer
32 views

How to precompute one sequence in a sequence-pair task when using BERT?

BERT uses separator tokens ([SEP]) to input two sequences for a sequence-pair task. If I understand the BERT architecture correctly, attention is applied to all inputs thus coupling the two sequences ...
user avatar
0 votes
2 answers
51 views

validation/test set uniqueness question

Hopefully a simple question, but it's a little unclear to me on how best to separate train/validate/test sets. I have say 100 examples of class A. I'm classifying text into either class A, which I ...
user avatar
  • 115
1 vote
0 answers
49 views

Variable batch size for inputs of different length

We're fine-tuning a GPT-2 model (using the Adam optimizer) to some posts from a social network. The length of these posts varies quite dramatically, so while some are only two tokens long, others can ...
user avatar
2 votes
1 answer
43 views

Why tfidf of one document is not zero?

I'm new to nlp. Recently I wanted to do little nlp tasks, and faced strange thing. That is I have run the following code ...
user avatar
  • 123
0 votes
0 answers
17 views

In a CWI task, does POS number order matter?

In a task where we need to identify which words are considered difficult for a given sentence, one important feature is part of speech tagging. This can be done by ...
user avatar
0 votes
1 answer
43 views

Word-level text generation with word embeddings – outputting a word vector instead of a probability distribution

I am currently researching the topic of text generation for my university project. I decided (ofc) to go with a RNN getting a sequence of tokens as input with a target of predicting the next token ...
user avatar
  • 101
1 vote
1 answer
86 views

How to evaluate triple extraction in NLP?

I am current NLP work, I am extracting triples using triple extraction function in Stanford NLP and Spacy libraries. I am looking for a good method to evaluate how good the extraction has been? Any ...
user avatar
  • 133
2 votes
1 answer
69 views

Encode a set of skills into a feature

I am working with a dataset where users have a set of skills. I have more than 500 skills and I was wondering what is the best way of encoding a vector, e.g., ...
user avatar
  • 143
0 votes
1 answer
25 views

Issue with sarcasm detection

I am working on the Reddit dataset for sarcasm detection but the sarcastic data points(1) are showing zero percent recall, precision, and accuracy however nonsarcastic are showing 100% recall and 50 ...
user avatar
0 votes
0 answers
31 views

NLP model to fill in the blanks given a document

Let's say that I have a document that has sentences containing information about my first name, last name, place of residency, car, salary, and age. Example: "At the age of 29, John Kean managed ...
user avatar
  • 101
0 votes
0 answers
49 views

Text classification length

I have a set of text examples I need to learn as class A, and they are of varying lengths, say 10 sentences to 1 sentence long. I have to parse a document to find those strings of text that match one ...
user avatar
  • 115
1 vote
1 answer
56 views

Is binary classification the right choice in this case?

I am somewhat new to text classification and I have some questions if you folks can help: I have some text I need to be able to classify as belonging to a single class or not (usually 1-10 sentences ...
user avatar
  • 115
0 votes
0 answers
9 views

Need some ideas for specified ML task

So I got many boxes and every box may contains many items (from 1 to 50), for example: Box1 : ball, small ball, table, table Box2 : golden ball Box3 : tea Box4:: t-shirt, t-shirt2 ... For chosen box I ...
user avatar
2 votes
1 answer
455 views

What is the difference between TextVectorization and Tokenizer?

What is the difference between the layers.TextVectorization() and ...
user avatar
0 votes
0 answers
18 views

How did AILabs Team get such performance in the superglue benchmark?

If we look at superglue (https://super.gluebenchmark.com/leaderboard) benchmark leaderboard, it may seem that AILabs does not perform super well. But if we look at the model card (https://super....
user avatar
  • 1
1 vote
0 answers
40 views

how to extract the following keywords containing cells from all columns of csv dataset and copy to new Analysis column

...
user avatar
  • 11
0 votes
0 answers
142 views

How to perform entity level train-val-test split for NER task?

A normal and stratified split option is provided by sklearn method that can be used for ML problems like multi-class classification. This is relatively easier to do as (1) one sample has one class, ...
user avatar
  • 101
1 vote
1 answer
14 views

Is there a way to train Doc2Vec on a corpus of docs and be able to take a novel doc and see how similar it is to the trained corpus?

I have a project idea, where I train a bunch of documents on Doc2Vec and then take a novel, input doc, and ideally be able to be told how similar it is to the docs supplied for training as a whole or ...
user avatar
  • 133
1 vote
1 answer
59 views

When to use GloVe vocabulary vs. building a vocabulary from the training data?

While studying some (pytorch) examples that use pretrained GloVe vectors I came across two variants: Use the vocabulary of the GloVe vectors and thus initialize the embedding layer with the ...
user avatar
1 vote
0 answers
12 views

What are the best ways to convert different parts of speech to noun?

I am working on a topic modelling task. I want to convert different parts of speech such as adjectives, verbs to noun. What are the best ways of doing this? I have tried lemmatization using NLTK ...
user avatar
0 votes
0 answers
36 views

Imbalanced NLP text classification

I'm trying to solve a multi-class text classification task with 3 classes. I have an initial pretty balanced but small dataset. When I start to mine additional data I can't always find a lot of new ...
user avatar
  • 157
1 vote
1 answer
45 views

Softmax in Sentiment analysis

I am going to do Sentiment Analysis over some tweet texts. So, in summary we have three classes: Positive, Neutral, Negative. If I apply Softmax in the last layer, I will have the probability for each ...
user avatar
1 vote
1 answer
24 views

Questions of understanding - Fast Lexically Constrained Decoding with Dynamic Beam Allocation for Neural Machine Translation

I'm currently analysing the paper Fast Lexically Constrained Decoding with Dynamic Beam Allocation for Neural Machine Translation (Post, Vilar 2018): https://arxiv.org/abs/1804.06609 I have ...
user avatar
4 votes
1 answer
550 views

BERT vs GPT architectural, conceptual and implemetational differences

In the BERT paper, I learnt that BERT is encoder-only model, that is it involves only transformer encoder blocks. In the GPT paper, I learnt that GPT is decoder-only model, that is it involves only ...
user avatar
  • 165
0 votes
0 answers
13 views

What should I visualize for humor detection model to gain some useful insight?

I was going through bunch (1,2,3) of humor detection paper. But most papers don't include any visualizations, say some graph related to model being trained. I was thinking to train some language ...
user avatar
  • 137
1 vote
2 answers
25 views

Automatically finding business opportunities in text documents

I am new to machine learning and NLP. I am exploring the possibility of using one of these approaches to automatically examine a large collection of text documents and determine, first of all, if they ...
user avatar
  • 11

1
3 4
5
6 7
47