Reddit Ner Place Names

🚀 Reddit NER for place names

A fine - tuned bert - base - uncased model for extracting place names from social media text.

This model is a fine - tuned bert - base - uncased for named entity recognition. It was trained using the wnut_17 dataset along with 498 additional comments from Reddit. Its sole purpose is to extract place names from social media text, and other entities have been removed. The model was developed with two main objectives:

1. To achieve better NER results on social media.
2. To focus only on place names.

🚀 Quick Start

Basic Usage

from transformers import pipeline

generator = pipeline(
    task="ner",
    model="cjber/reddit-ner-place_names",
    tokenizer="cjber/reddit-ner-place_names",
    aggregation_strategy="first",
)

out = generator("I like reading books. I live in Reading.")

The output out is:

[
    {
        "entity_group": "location",
        "score": 0.94123614,
        "word": "reading",
        "start": 32,
        "end": 39,
    }
]

✨ Features

• Improved NER on Social Media: The model is trained to perform better in the context of social media text.
• Place Name Focus: It specifically targets place names, ignoring other entities.
• Metonymy Handling: In theory, it can detect and ignore metonyms. For example, in the sentence Manchester played Liverpool last night in Liverpool., it correctly identifies only the actual location mention:

[
    {
        "entity_group": "location",
        "score": 0.9975672,
        "word": "liverpool",
        "start": 42,
        "end": 51,
    }
]

📦 Installation

No specific installation steps are provided in the original document.

💻 Usage Examples

Basic Usage

from transformers import pipeline

generator = pipeline(
    task="ner",
    model="cjber/reddit-ner-place_names",
    tokenizer="cjber/reddit-ner-place_names",
    aggregation_strategy="first",
)

out = generator("I like reading books. I live in Reading.")

The output out is:

[
    {
        "entity_group": "location",
        "score": 0.94123614,
        "word": "reading",
        "start": 32,
        "end": 39,
    }
]

Metonymy Example

In the sentence Manchester played Liverpool last night in Liverpool., the model outputs:

[
    {
        "entity_group": "location",
        "score": 0.9975672,
        "word": "liverpool",
        "start": 42,
        "end": 51,
    }
]

📚 Documentation

Model Code

For the model code, please visit the Model GitHub Repository.

📄 License

This model is released under the MIT license.

📦 Dataset and Metrics

🚀 Widget Examples

• Metonyms: "Manchester played Liverpool last night in Liverpool."
• Slang / informal text: "i live in brum - slang for birmingham"