Showcases two POS tagging models trained using a universal tagset.
The Most Frequent Class tagger simply selects the POS tag that appears most often for each word. This received a 93% accuracy rate on the test data.
The Hidden Markov Model tagger considers each tag to be a state in a Finite State Machine, alongside the start and end states. A set of probabilities allowing transition between the states is used to account for the order that tags tend to be found in natural language. Each state also has a set of emission probabilities which determine how likely each word is to be emitted by that state. This received a 95% accuracy rate on test data.
It should be noted that the accuracy would decrease on larger tagsets, as there will be more room for error. These accuracy rates are slightly inflated due to the universal tagset used.
The original GitHub repo for this project can be found here.
In this notebook, you'll use the Pomegranate library to build a hidden Markov model for part of speech tagging with a universal tagset. Hidden Markov models have been able to achieve >96% tag accuracy with larger tagsets on realistic text corpora. Hidden Markov models have also been used for speech recognition and speech generation, machine translation, gene recognition for bioinformatics, and human gesture recognition for computer vision, and more.
The notebook already contains some code to get you started. You only need to add some new functionality in the areas indicated to complete the project; you will not need to modify the included code beyond what is requested. Sections that begin with 'IMPLEMENTATION' in the header indicate that you must provide code in the block that follows. Instructions will be provided for each section, and the specifics of the implementation are marked in the code block with a 'TODO' statement. Please be sure to read the instructions carefully!
You can choose one of two ways to complete the project. The first method is to use the Workspace embedded in the classroom in the next lesson. The Workspace has already been configured with all the required project files for you to complete the project. Simply open the lesson, complete the sections indicated in the Jupyter notebook, and then click the "submit project" button.
NOTE: If you are prompted to select a kernel when you launch a notebook, choose the Python 3 kernel.
Alternatively, you can download a copy of the project from GitHub here and then run a Jupyter server locally with Anaconda.
NOTES: These steps are not required if you are using the project Workspace.
-
(Optional) The provided code includes a function for drawing the network graph that depends on GraphViz. You must manually install the GraphViz executable for your OS before the steps below or the drawing function will not work.
-
Open a terminal and clone the project repository:
$ git clone https://github.com/udacity/hmm-tagger
- Switch to the project folder and create a conda environment (note: you must already have Anaconda installed):
$ cd hmm-tagger
hmm-tagger/ $ conda env create -f hmm-tagger.yaml
- Activate the conda environment, then run the jupyter notebook server. (Note: windows users should run
activate hmm-tagger)
hmm-tagger/ $ source activate hmm-tagger
(hmm-tagger) hmm-tagger/ $ jupyter notebook
Depending on your system settings, Jupyter will either open a browser window, or the terminal will print a URL with a security token. If the terminal prints a URL, simply copy the URL and paste it into a browser window to load the Jupyter browser. Once you load the Jupyter browser, select the project notebook (HMM tagger.ipynb) and follow the instructions inside to complete the project.
See below for project submission instructions.
Your project will be reviewed by a Udacity reviewer against the project rubric here. Review this rubric thoroughly, and self-evaluate your project before submission. All criteria found in the rubric must meet specifications for you to pass.
Once you have completed all of the code implementations, you need to finalize your work by exporting the iPython Notebook as an HTML document. Before exporting the notebook to html, all of the code cells need to have been run so that reviewers can see the final implementation and output. You must then export the notebook by running the last cell in the notebook, or by using the menu above and navigating to File -> Download as -> HTML (.html) Your submissions should include both the html and ipynb files.
Add the "hmm tagger.ipynb" and "hmm tagger.html" files to a zip archive and submit it with the button below. (NOTE: If you complete the project in the workspace, then you can submit directly using the "submit" button in the workspace.)