How to measure table and page content length
- Applies to:
- MindTouch (current)
- Role required:
- Admin
Having a broad view of content indicates how much content (text length) is on each article, which articles contain tables, and the dimensions of those table.
API endpoint:/@api/deki/site/content/report
Logs are updated at midnight PST every day and are delivered in CSV format. All Seated users can access the table and page content length report.
Table and page content length report details
Field name | Description | Example |
PAGE_ID | The ID of the page | 23 |
PAGE_CHAR_COUNT | How many characters are on the page | 547 |
PAGE_URI | The URI of the page | https://[site].com/Template:Custom |
CONTAINS_TABLE | Whether or not a page contains a table |
FALSE TRUE |
TABLE_CELLS_COUNT | How many cells are in the table | 12 |
NAMESPACE | The namespace the page is found in |
MAIN TEMPLATE |
IS_DRAFT | Whether or not the page is a draft |
FALSE TRUE |
How large can tables be?
Table size estimations by content type
Tables size may vary depending on content:
Table Type | Example | Average Words per Cell | Average Characters per Cell |
Short Table | Names, Numbers (for example, "John," "35") | 1-3 | 5-15 |
Medium Table | Descriptions, Full or partial phrases (for example, "Software Engineer," "Product Management," "Gulf of Mexico" | 2-5 | 15-40 |
Long Table | Rich content, Explanations (for example, "Experienced AI researcher with a focus on LLMs and Knowledge Management") | Full sentences | 40-200 |
Maximum table sizes
To indicate structure to an LLM, it is critical to preserve table content in its entirety (including HTML tags). While a human reader may visually understand tabular data, an LLM does much better with the raw HTML to inform responses. Tables are embedded with markup to maintain their structure, including elements such as <td>
, <th>
, <table>
, <ul>
, <ol>
, <li>
, <a>
, <br>
, and relevant attributes like scope. Styling, classes or <span>
elements are not included.
With a short table, there is less textual content, so you can have more cells. In long tables with content-heavy descriptions, sentences have less room for cells. Use the following guidelines for max table cells based on table type (short, medium, long):
Kernel limit | Short table | Medium table | Long table |
10 | ~2,222 cells | ~1,020 cells | ~248 cells |
5 | ~4,444 cells | ~2,040 cells | ~496 cells |
Use the CSV data
Reduce high cell area tables
After downloading the CSV report, filter table cell counts in descending order to find your largest tables. If your largest tables exceed the recommended table size, refine them:
- Reduce unnecessary columns or rows.
- Break tables into smaller tables.
- Add a succinct description of the tables purpose and content in a sentence or two above the table.
- Remove nested tables.
- Do not use tables as a styling mechanism. Use CSS from Control Panel instead.