# Models
## Works
### Field Frequency
The existence of certain fields were found to vary by record. To understand the impacts,
the following is a frequency distribution over 10 million records:
| created | 100% |
| last_modified | 100% |
| latest_revision | 100% |
| key | 100% |
| type | 100% |
| revision | 100% |
| title | 99.99% |
| authors | 95.23% |
| subjects | 58.8% |
| covers | 32.0% |
| subject_places | 24.7% |
| first_publish_date | 18.57% |
| id | 13.47% |
| lc_classifications | 9.9% |
| subject_people | 7.7 |
| subject_times | 6.07 |
| dewey_number | 4.8% |
| description | 4.4% |
| subtitle | 2.4% |
| links | 0.19% |
| first_sentence | 0.3% |
| excerpts | 0.2% |
| cover_edition | 0.008% |
| location | 0.0002% |
This client is choosing not to support fields with less than 20% frequency until there is an explicit
reason to do so.
### Data Quality Issues
1. There is one record in the Data Dump (`OL25303131W`) that doesn't have a title. However,
this seems to be "fixed" via the API responses.
2. There are a number of fields that only 1 record has:
1. `original_languages`
2. `table_of_contents`
3. `number_of_editions`
4. `notes`
5. `notifications`
3. The `description` field has inconsistent data format.
**Example**:
| Identifier | Data Type | Value |
|------------|-----------|-----------------------------------------|
| OL499789W | object | { "type": "/type/text", "value" "..." } |
| OL39625W | string | "..." |
4. The `type` within the `authors` array has an inconsistent data format
**Example**:
| Identifier | Data Type | Value |
|------------|-----------|-----------------------------------------|
| OL499789W | object | { "type": "/type/text", "value" "..." } |
| OL39625W | string | "..." |