Stack Exchange Network

Stack Exchange network consists of 174 Q&A communities including Stack Overflow, the largest, most trusted online community for developers to learn, share their knowledge, and build their careers.

Visit Stack Exchange

The tag has no usage guidance.

0
votes
0answers
12 views

Using clickthrough data for training a ranking function

Suppose that you have clickthrough data in the following form -- (query, clicked url, frequency). I wonder if there is any way of using the data to train a ranking function. Naively, you can treat ...
0
votes
0answers
19 views

Estimating a surprise of a word in context

What will be the best way to estimate the entropy/surprise of a word in a specific context? Let's say to compare the surprise of: context: "I watched the movie in my" word: Computer I ...
0
votes
1answer
35 views

How to compute k-means on provided inverted index using tf-idf

I built an inverted index to represent the following sample documents: ...
0
votes
0answers
12 views

Which curve comparison should I use to evaluate the performance of a recommender?

I am building a recommender system on the Last.FM dataset (link here) (1,892 users and 17,632 artists and the number of times a particular artist was listened to by a user). Next, the raw dataset was ...
0
votes
0answers
15 views

Positional similarity

What is a good positional similarity measure? Bottom line - words, which are next to each other should be ranked higher (no semantic or fuzzy matching needed). I want to use this in addition to TF/IDF-...
1
vote
2answers
49 views

What are commonly used methods to represent a document by a vector?

Methods that I know of Bag of words + weighting: tf-idf, bm25 Topic models: LSA, LDA Word/sentence/document embedding Are there other commonly used methods to represent a document by a vector?
0
votes
0answers
24 views

Comparing different ranking methods' results

I have been looking for literature about comparing the retrieval results of two methods. One method results in a list of ranks whereas the second method results in a list of sets of ranks. Do you have ...
0
votes
0answers
24 views

How to combine heterogeneous image features extracted with different algorithms for similar image retrieval?

Say I have access to several pre-trained CNNs (e.g. AlexNet, VGG, GoogleLeNet, ResNet, DenseNet, etc.) which I can use to extract features from an image by saving the activations of some hidden layer ...
1
vote
0answers
21 views

ML Approach for Optimizing Field Boosts for Search (Information Retrieval)

I have been experimenting with different methods for tuning our search engine's field boosts. In Solr or Elastic Search, you specify the importance of matches in each field when configuring the search ...
0
votes
0answers
15 views

metric score when ranking information is only available at test time

I am training a machine learning model where each training sample consists of a set of "competitors" and the aim is to predict the winner (there might be more than one winner. I don't care about ...
0
votes
0answers
46 views

Clustering huge amount of vectors

I have a huge amount of vectors (about 2 millions). Vectors may be different by size (length ranges from 1 to 86K), each element of a vector is a number and the elements are sorted. (E.g. v_1=[1,3,4,5,...
2
votes
1answer
445 views

Remove bias in ranking evaluation

In ranking, a commonly used method to evaluate the performance of a ranking algorithm is calculating ranking metrics such as MAP, NDCG. In use cases where there is no ground truth signal, some proxy ...
0
votes
0answers
27 views

Multi-class classification with different number and categories of class for each datum

I have a problem that is a bit like multiclass classification. There is only one correct classification, but the number of possible classifications (all of which baring 1) is different for each thing ...
2
votes
1answer
53 views

inverting the binomial distribution: probability distribution for number of trials necessary to have a given number of successes

The binomial distribution gives me a distribution for the number of successes in several Bernoulli trials, k, given parameters N the total number of trials and q, the success probability for one ...
0
votes
0answers
56 views

Information retrieval Measures of association/correlation

I try to understand correlation measures of terms. Therefore I found the following exercise: Which term is more correlated/associated with the term pool: ...
1
vote
0answers
35 views

Learning to rank and traditional information retrieval evaluation

I have some questions about best practices in information retrieval (unsupervised) vs learning to rank evaluation. How necessary in a train-validation-test or cross-validation scenario? is it ...
0
votes
0answers
33 views

Average Precision. Ranking only one item as good as ranking two?

With respect to Average Precision(AP) for recommender systems, if a recommender gets two items in rank vs another recommender which only gets one item in rank, they would still have the same Average ...
1
vote
1answer
22 views

What contests/datasets for expertise retrieval ?

Expertise retrieval is a difficult task to define. This article describes the task of expert finding as "finding the right person with the appropriate skills and knowledge". Applications exist in ...
0
votes
1answer
382 views

Extracting Part of Speech (Source and Destinations) using text mining/NLP?

I need to extract the source and destination terms from the text documents using text mining/NLP/Information Retrieval ? ex : i am travelling from New York to London. i am heading towards ...
1
vote
0answers
235 views

NDCG for recommender algorithm

I need to apply NDCG over the results of a recommender algorithm, but was not able to find any proper example that suits my use case in order to find out if my implementation is correct. Here is my ...
0
votes
2answers
288 views

What algorithm should I use to predict a continuous dependent variable from multiple continuous & categorical independent variables?

I'm software engineer of an E-commerce company, facing a problem like this: An e-commerce shop sells their products daily and wants to know what conditions that might improve their sales. I'm ...
3
votes
1answer
2k views

Both validation loss and accuracy goes up in neural network

I'm training a 2-layer CNN model on audio samples, represented as CQT. There are ≈160k samples, many that are very similar since they originate from the same instrument and/or audio file. 10% have ...
4
votes
0answers
51 views

How can I deal with the mismatch between the vocabularies of questions and answers in a closed domain QA system?

I am building a question answering system that given a legal document attempts to answer questions related to the document. For example a tenancy agreement is given to the system and the user asks ...
2
votes
0answers
102 views

How to adjust your dataset so it fits a power-law? [closed]

I have created a dataset of pictures taken at the museum of different paintings. The dataset is divided into 113 different categories (paintings) and contains around 4.8k images. Just to be clear: ...
1
vote
0answers
91 views

Representing Text for Section Labeling While Avoiding Bias

Problem Background: I have free form text data with more or less arbitrary formatting/structure, but semantically it can be broken down like this: ...
2
votes
1answer
348 views

How to find “similar documents” after a Latent Dirichlet Allocation model is built

Let's say I run an LDA model with 3 topics on 5 documents. After the model is learned (with Gibbs sampling presumably), I have topic distribution for each document, shown as the following: My ...
0
votes
2answers
1k views

Drawbacks with Cosine Similarity

I am assessing the similarity between documents represented as vectors of tf-idf values. I know that the cosine similarity is a well-defined and commonly used measure in information retrieval. ...
1
vote
0answers
85 views

Why F-1 is a better score than the harmonic mean between true negative rate and recall?

Beside of the easy interpretation, F1 measure is very sensible to the relative frequency between positives and negatives. ROC AUC metric don't, but it is difficult to optimize. Are there any special ...
0
votes
0answers
84 views

What is a rotation matrix and how to implemnt it?

In Revisiting the VLAD Image Representation the authors introduce Local Coordinate System, i.e. they: we learn off-line (for each visual word) a rotation matrix Qi from training descriptors mapped ...
1
vote
0answers
353 views

Union of good feature sets degrades accuracy

I am doing binary text classification and I have some feature sets (unigrams, bigrams, dependencies, etc.) and each one of these performs very good individually. For example unigrams alone achieve 89% ...
1
vote
0answers
92 views

Contextual Matching algorithm implementation in Python

I have been trying to implement an algorithm using Python in order to perform contextual matching in a set of documents. My ultimate goal would be to be able to perform queries using positive keywords ...
1
vote
0answers
24 views

Average Number of Documents Containing a Term

I am working on a research project in which I am using inverted index for terms and documents in one of the techniques. I am doing algorithmic analysis for all the methods so that I can compare and ...
3
votes
2answers
69 views

Is there a ranking metric based on percentages that favors larger magnitudes?

I have two groups, "in" and "out," and item categories that can be split up among the groups. For example, I can have item category A that is 99% "in" and 1% "out," and item B that is 98% "in" and 2% "...
1
vote
2answers
496 views

Can we use Bag of Visual Words to compute similarity between images directly?

I'm implementing a Content Based Image Retrieval application (CBIR). I've read about the Bag of Features model and it's considered an intermediate-step algorithm in some application. For example, ...
2
votes
0answers
74 views

F-measure and hypothesis testing

I would compare two classifiers (A and B), where B is obtained from A. I would exploit the F-measure computed on two samples (of two independent populations, p1 and p2, respectively): A -> F-...
5
votes
1answer
2k views

Difference between Log Entropy Model and TF-IDF Model?

I would like to understand what are the differences/advantages in using TF-IDF or the Log Entropy model for represeting documents and queries in an information retrieval system using diferent weights. ...
2
votes
1answer
720 views

Where did sublinear tf-idf originate?

I have often come across this weighting scheme for tf-idf (term frequency - inverse document frequency) in text mining. I am wondering where it came from (for citations). I've searched very rigorously,...
0
votes
1answer
31 views

Where do initial document values come from in K-means document clustering?

So the K-means algorithm seems simple enough as I understand it: given some documents, turn those documents into points, initialize some number of k (centroids), assign document-points to nearest ...
1
vote
0answers
48 views

How to decide what is the relevant group in a precision and recall computation?

One of the most famous measurements for an information retrieval system is to compute its precision and recall. For both cases, we need to compute the number of total relevant documents and compare it ...
0
votes
0answers
200 views

topic modeling for image retrieval

I'm interested in learning a topic model from a bag of visual words for image retrieval. I can compute V cluster centers (visual words) of SIFT descriptors at keypoints for each training image and fit,...
7
votes
1answer
2k views

The Effect of Stopword Filtering prior to Word Embedding Training

Recently I have played with the pretrained GLOVE word embedding model for Twitter http://nlp.stanford.edu/projects/glove/ I notice that common stopwords are existing in the model. That is, there is ...
1
vote
1answer
595 views

how to find the nearest neighbor of a sparse vector

I have about 500 vectors,each vector is a 1500-dimension vector, and almost every vector is very sparse-- I mean only about 30-70 dimension of the vector is not 0。 Now, the problom is that here is a ...
2
votes
0answers
444 views

Is there an algorithm for determining scoring function (Utility function) in ranking instances?

So while going through the topic of Preference learning, I came to know about "instance ranking". Since the problem which I'm working on requires me to rank the instances (data point), is there any ...
1
vote
0answers
191 views

Is this ordering of examples redundant in Rank SVM?

I was studying RankSVM[1] and I was observing the ranking pairs listed in the example here. Formulation can be seen on page 4 of this paper. One of the ordering is:...
3
votes
1answer
488 views

Normalised score for BM25

BM25 provides a function that assigns a score that is a function of a query and a document. The score is computed for each document in the collection and can be used to rank documents, but this score ...
2
votes
1answer
89 views

Explaining the steps of a visualization tool

this is my first post on CrossValidated. I've done, for academic purpose, a web tool doing this process: web scraping from various sites pre-process the responses (cleaning, error and redundance ...
1
vote
0answers
1k views

Typical range of values for TFIDF

I am working on a text corpus. Each line contains between 10 and 50 words. There are around 25 000 words in the whole text and 1 000 000 lines. I turned this corpus into its tf-idf representation. I ...
8
votes
1answer
376 views

Why does Lucene IDF have a seemingly additional +1?

From the Lucene docs $\text{IDF} = 1 + \log\left(\frac{\text{numDocs}}{\text{docFreq}+1}\right)$ In other references (i.e. wikipedia), IDF is typically calculated as $\log\left(\frac{\text{numDocs}}{...
1
vote
1answer
1k views

Average precision when not all the relevant documents are found

I can't find on the Internet a proper source that explains this. I have built a search engine that for a particular query retrieves 5 relevant document out of the 10 relevant documents. When I ...
6
votes
4answers
5k views

What is the best algorithm to find similar text documents?

I have many text documents and I would like to find similar documents to each document within my data set. Is Latent Dirichlet Allocation (LDA) the best way to do that, or are there other algorithms ...