Main Page

WikiFashion WikiBase

Scope of the project

The WikiFashion WikiBase project stores information about fashion designers, fashion brands and the relative collections. By mapping the history of recent fashion in a WikiBase format we aim to make research on fashion creativity and innovation easier. Data comes from Wikidata, Wikipedia, Vogue and Fashion Model Directory (FMD) and provenance of each information will be clearly mapped through the reference tool. The source code that was used to populate this knowledge base can be found in LINK GITHUB.

Example queries

Education information with start and end date

PREFIX wbt: <https://wikifashion.wikibase.cloud/prop/direct/>
PREFIX wb: <https://wikifashion.wikibase.cloud/entity/>
PREFIX pq: <https://wikifashion.wikibase.cloud/prop/qualifier/>  
PREFIX ps: <https://wikifashion.wikibase.cloud/prop/statement/> 
PREFIX p: <https://wikifashion.wikibase.cloud/prop/>            

SELECT ?fashiondesignerLabel ?schoolLabel ?startDate ?endDate WHERE {
  ?fashiondesigner p:P113 ?statement.  
  ?statement ps:P113 ?school. 
  
  OPTIONAL {
    ?statement pq:P119 ?startDate.  
    ?statement pq:P120 ?endDate.    
  }
  
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}

Employment information with start and end date

PREFIX wbt: <https://wikifashion.wikibase.cloud/prop/direct/>
PREFIX wb: <https://wikifashion.wikibase.cloud/entity/>
PREFIX pq: <https://wikifashion.wikibase.cloud/prop/qualifier/>  
PREFIX ps: <https://wikifashion.wikibase.cloud/prop/statement/>  
PREFIX p: <https://wikifashion.wikibase.cloud/prop/>            

SELECT ?fashiondesignerLabel ?employerLabel ?startDate ?endDate WHERE {
  ?fashiondesigner p:P114 ?statement.  
  ?statement ps:P114 ?employer.  
  
  OPTIONAL {
    ?statement pq:P119 ?startDate.  
    ?statement pq:P120 ?endDate.    
  }
  
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}

How is the information extracted from text?

import ollama
import instructor
import openai
from pydantic import BaseModel
from typing import List, Optional

class FashionDesigner(BaseModel):
    educated_at: Optional[List[list]] = None  
    employer: Optional[List[list]] = None    
    work_location: Optional[List[list]] = None 
    award_received: Optional[List[list]] = None  


def send_chat_prompt_instructor(prompt, model):
    client = instructor.from_openai(
        openai.OpenAI(
            base_url="http://localhost:11434/v1" if not "gpt" in model else None,
            api_key= "ollama" if not "gpt" in model else openai_api_key),
        mode=instructor.Mode.JSON,)
    resp = client.chat.completions.create(
        model=model,
        temperature = 0.5,
        messages=[
                {"role": "system", "content": "You extract structured information from biographies about a fashion designer's education, employment, work locations, and awards. Only generate the output in JSON format."},  
                {"role": "user", "content": prompt}],
        response_model= FashionDesigner,)
    response = (resp.model_dump_json(indent=2))
    return response

def main_generate_instructor(prompt,model):
    response = send_chat_prompt_instructor(prompt,model)
    return response

def prompt_template(biography_example, biography_KG_example, biography_test, ontology = None, concepts = None, school = None, houses_designer = None):
    if ontology is None:
        basic_instruction = """Extract knowledge triples from the text.In the output, only include the triples in the given output format."""
    else:
        basic_instruction = f"""Given the following ontology and sentences, please extract the triples from the sentence according to the relations in the ontology. In the output, only include the triples in the given output format.
        {concepts}
        {ontology}"""
        
    if school is None:
        entities_from_text_example = None
        entities_from_text_test = None
    else:
        entities_from_text_example = f"""Entities of type academic institution in Example Sentence {find_names_in_text(school, biography_example)}
        \nEntities of type fashion house in Example Sentence {find_names_in_text(houses_designer, biography_example)}"""
        entities_from_text_test = f"""Entities of type academic institution in Test Sentence {find_names_in_text(school, biography_test)}
        \nEntities of type fashion house in Test Sentence {find_names_in_text(houses_designer, biography_test)}"""

    prompt_ont =f"""
    {basic_instruction}
    \nExample Sentence: {biography_example}
    {entities_from_text_example}
    \nExample Output:  {biography_KG_example}
    \nTest Sentence: {biography_test}
    {entities_from_text_test}
    \nTest Output:
    """
    return prompt_ont

How is information structured?

Data is stored in Wikibase in the shape of Items. Each item is accorded its own page. Items are used to represent all the things in human knowledge, including topics, concepts, and objects. In the WikiFashion Wikibase all the items are related to fashion and the history of fashion and of the fashion designers. In this Wikibase you can find entities for the fashion designer, countries and places where the fashion designers lived and worked, academic institutions and fashion houses. For example, the "1988 Summer Olympics", "love", "Elvis Presley", and "gorilla" can all be items. Items are made up of Statements that describe detailed characteristics of an Item. A statement (graph format: Subject-Predicate-Object) is how the information we know about an item - the data we have about it - gets recorded in your Wikibase instance. This happens by pairing a property with at least one value; this pair is at the heart of a statement. Statements also serve to connect items to each other, resulting in a linked data structure.

Check out this visualization of the linked data structure

The property in a statement describes the data value, and can be thought of as a category of data like "color", "population," or "Commons media" (files hosted on Wikimedia Commons). The value in the statement is the actual piece of data that describes the item. Each property has a data type which defines the kind of values allowed in statements with that property. For example, the property “date of birth” will only accept data in the format of a date.

Check out this visualisation of the structure of an item

Example In order to record information about the occupation of Marie Curie, you would need to add a statement to the item for Marie Curie (Q7186). Using the property, occupation (P106), you could then add the value physicist (Q169470). You could also add the value chemist (Q593644). Note how both chemist and physicist are each their own item, thereby allowing Marie Curie to be linked to these items.

How to create items + properties + (what is a good property)

Create a new Item with Special:NewItem on the menu to the left. You will be taken to a page that asks you to give the new item a label and a description. When you're done, click "Create".

Create the property with Special:NewProperty. Property entities can be edited like item entities with labels, descriptions, aliases, and statements. Property labels should be as unambiguous as possible so that it is clear to a user which property is the correct one to use when editing items and adding statements. Properties rarely refer to commonly known concepts but they are more constructs of the Wikidata with specific meanings. Unlike items, property labels must be unique. Property descriptions are less relevant for disambiguation but they should provide enough information about the scope and context of the property so that users understand appropriate usage of the property without having to consult additional help. Property aliases should include all alternative ways of referring to the property.

Example: property: P161 label: cast member description: actor performing live for a camera or audience aliases: film starring; actor; actress; starring

To create and delete data using tools, have a look at this list in the documentation.

Another way is via the Wikibase API that allows querying, adding, removing and editing information on Wikidata or any other Wikibase instance. A new version of the API is in the works with the REST API.

Don’t panic -> links to help

This is a lot of information and possibilities and might seem daunting. Don’t panic. There’s a lot of documentation for you to read, many other Wikibases you can look at to get inspired, a community that is happy to help out with any questions, and the development team who’s happy to support you where needed.

Last but not least

To edit this page, make sure you are logged in and click on -edit- on the top right of this page to change or remove the content.