Should you use a dictionary or an ontology?
When you are creating a project, you will need to decide what project results you are looking to obtain. This choice will determine the path in which your data will be analyzed.
Dictionaries are useful for detecting keywords or patterns in your data. A common use cases for a dictionary is content moderation (detecting sensitive or offensive content).
Ontologies are used to structure you data. For example, ontologies can be used to extract information to fill out databases and CRMs.
In this use case, we are going to take a look at a shipping and delivery company that is looking to improve customer relations and obtain sales objectives. This same shipping and delivery company would like to automize their database entries for their product information. We will take a look at both of these projects and determine whether a dictionary or ontology would be appropriate.
Dictionary use case
For the use case of the shipping company that pertains to improving customer relations and reaching sales objects, this can be achieved by using a dictionary. By using a dictionary you can answer the following questions:
- Are the sales representatives proposing the ideal services and/or products?
- Are customer questions being answered accurately?
- Is the sales representatives using appropriate language?
- Is your online community following your guidelines?
Dictionaries are great for content moderation and can easily flag insults. To start your moderation you can create your own diction like the one below by adding your words manually or uploading your dictionary in .csv format.
Dictionaries allow you to detect patterns and key terms to help you to make key decisions regarding the analysis of your data. You can use dictionaries to extract key terms and patterns from you data such as:
- Phone numbers, siren number, zip codes, etc.
- Company names (useful for competitive analysis)
Ontology use case
In the use case of the shipping company, we will look at how ontologies can help to extracting information for a database or CRM. Ontology allow you to assign properties to classes. Think of an identity card, the class is the main item and the information relating to the class are the properties. For the shipping company use case we are looking to add data to a database for each product and possible properties.
For a classic package we will look for the corresponding information: weight, price, destination, sender, quantity, additional services, etc.
We can look for related information for each product. This will help to train and build a model that will automize this task for building you database entries.
This process can also be applied for adding data to a CRM.
- The customer can be the class and the properties relating to that customer can include: shipping orders, shipping destinations, subsidiaries etc.
🤔 Still have questions? Check out our FAQ for ontology