Introduction
Dumpor is a program that takes a list of title-topic-section triplets and returns the same list with topics extracted from section content using machine learning. The training data contains sections that have been manually classified into the intended topics.
dumpor uses machine learning to read your section content and make an educated guess as to how best to classify it into a topic. The available topics are defined as they appear in training data, and depend on the training set used but common topics include “`Health, Lifestyle, Home & Garden“`.
dumpor is a program that takes a list of title-topic-section triplets and returns the same list with topics extracted from section content using machine learning.
We developed dumpor, a program that takes a list of title-topic-section triplets and returns the same list with topics extracted from section content using machine learning. The authors have already published their findings on this work in [1].
Dumpor is written in Python and uses scikit-learn to extract topics from text data. It’s easy to use, so if you’re interested in trying it out yourself (or if your company needs help with extracting specific topic sets), check out our repository!
The training data contains sections that have been manually classified into the intended topics.
The training data contains sections that have been manually classified into the intended topics. These are called “labels”. The model is then trained on these labels, so that it can learn how to classify new documents based on their content.
The label for each topic should be good enough to distinguish between similar topics in order to ensure that the model understands what you want it to understand. In other words, if a document has many different topics but still has some overlap between them (for example: “Health” and “Fitness”), then this could cause confusion when using your model later on—so make sure you don’t train with such data!
dumpor uses machine learning to read your section content and make an educated guess as to how best to classify it into a topic.
Dumpor uses machine learning to read your section content and make an educated guess as to how best to classify it into a topic.
For example, if you have an article about “cars,” the first thing that dumpor will do is parse through all of the words in your text for occurrences of words like “car” or “automobile.” If it finds enough matches, then it will create one or more topics for each match (for example: cars).
The available topics are defined as they appear in training data, and depend on the training set used, but common topics include “`Health, Lifestyle, Home & Garden“`.
The available topics are defined as they appear in training data, and depend on the training set used. A common topic might be “`Health, Lifestyle, Home & Garden“`.
The available topics are defined as they appear in training data, and depend on the training set used. A common topic might be “`Health (1), Lifestyle (2), Home & Garden (3)“`.
dumpor gives you suggestions for how to classify your blog posts by topic.
Dumpor is a tool for classifying blog posts by topic.
Dumpor uses machine learning to classify blog posts by topic, using the following steps:
- A training set of 1,000 most recent blog posts is used to train an algorithm that determines how well each post is classified as a “topic”.
- The algorithm then generates a list of title-topic-section triplets (a list of words), followed by their probabilities (the probability that each word occurs in that order). For example: if a user types “dog”, “cat” and “knows”, then we can say there’s a 60% chance that this user wrote about dogs or cats!
- If you want to check if your site has been classified correctly – open up the page where you have your blog published and look at all those numbers!
Conclusion
A part of this section is for you to decide on how much training data you’ll need to get good results. But if it’s not enough, we can always help!
