This project seeks to detect programming languages using datasets from Stack Overflow and Reddit.
The goal of the project is to make a model that could be used generally but it will specifically be used as a Named Entity Recognition exercise on Stack Overflow/Reddit with verticals for Sentiment Analysis. It is an attempt to investigate how the different communities think about different programming languages.
Typically, two models are evaluated using this project; a pattern matching model and a spaCy NER model to pursue a comparison between rule-based Statistics and Machine Learning approaches for NLP.
To export and run this workflow on your local machine, use the spacy project run package command.
The project.yml defines the data assets required by the
project, as well as the available commands and workflows.
The following commands are defined by the project. They
can be executed using spacy project run [name].
Commands are only re-run if their inputs have changed.
| Command | Description |
|---|---|
preprocess |
Convert the data to spaCy's binary format |
patternmod |
Generate a named entity recognition model based on rules. |
train |
Train a named entity recognition model |
evaluate |
Evaluate the model and export metrics |
package |
Package the trained model so it can be installed |
show-stats |
Show the statistics that compares both models. |
The following workflows are defined by the project. They
can be executed using spacy project run [name]
and will run the specified commands in order. Commands are only re-run if their
inputs have changed.
| Workflow | Steps |
|---|---|
all |
preprocess → patternmod → train → evaluate |
The following assets are defined by the project. They can
be fetched by running spacy project assets
in the project directory.
| File | Source | Description |
|---|---|---|
assets/stackoverflow-train.jsonl |
Local | JSONL-formatted training data |
assets/stackoverflow-valid.jsonl |
Local | JSONL-formatted validation data |
The following configuration files are defined by the project.
| File | Source | Description |
|---|---|---|
configs/config.cfg |
Local | CFG-formatted for base config |
configs/proglang_patterns.jsonl |
Local | JSONL-formatted rule patterns |
The following Python scripts are defined by the project.
| File | Source | Description |
|---|---|---|
scripts/preprocess.py |
Local | Pre-Processing Script |
scripts/save_pattern_model.py |
Local | Pattern NER Script |
scripts/print_stats.py |
Local | Results Comparison Script |
The commands and workflows can be used with the CLI as follows:
Initialize: project run
Command Execution: project preprocess
Workflow Execution: project all
Metrics: project show-stats
spaCy and spaCy Projects: Documentation
Explosion Templates: GitHub Repository
Part of this documentation has been auto-generated using the spacy project document command!
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent