Exact Match Term Frequency checks whether any entire line of a page’s body HTML is identical to one of the run’s keyword variations, then returns a log-transformed score: 1 + ln(count), where count is 1 if a line matches and 0 if none does. It is a compatibility measurement, not a real occurrence count — read it only alongside clearer keyword-usage factors.
| Factor ID | RG-KWD-007 |
|---|---|
| Family | Keyword Usage & Density |
| Measurement | Log-transformed binary line match |
| Measured zone | Raw body HTML without scripts |
What it measures
This compatibility factor tests whether a complete line of the body source equals one of the run’s keyword variations, then applies a natural-log term-frequency formula. It does not count how often a term appears; it asks a single yes-or-no question — does one whole line of the raw body HTML exactly equal a keyword variation — and expresses the answer through a logarithmic transform.
How RankGear measures it
RankGear builds an unescaped, case-insensitive expression from the sorted variation set and anchors it to an entire line. It sets the count to 1 when any line matches and otherwise to 0, then returns 1 + ln(count). Because the count is binary, the transform has only two possible outcomes.
count = 1 if any whole line of the raw body HTML == a keyword variation
count = 0 otherwise
score = 1 + ln(count)| Value | Meaning |
|---|---|
| 1 | A full body line exactly matched a keyword variation (count = 1, so 1 + ln(1) = 1). |
| -∞ | No line matched (count = 0, so 1 + ln(0) is mathematically negative infinity). |
How to optimize it
Do not optimize directly for the line-level implementation. Treat this value as an observation, not a target: use relevant variations naturally in visible content, and interpret this factor only alongside clearer keyword-usage factors. Reformatting a page so a keyword variation happens to occupy its own source line would satisfy the match without improving the page for readers.
Important considerations
- The numerator is binary, not a normal occurrence count — a page that repeats a variation many times scores the same as one that matches a single line once.
- A match produces 1; no match mathematically produces negative infinity, so the two outcomes are not on a smooth continuous scale.
- Source line breaks can change the result without changing the visible meaning of the page.
- Variation strings are not regex-escaped, for compatibility with the original implementation.
- Like every RankGear factor, this tracks a pattern within a measured result set. It correlates with position; it does not by itself make a page rank or guarantee a ranking.
Related factors
Part of the Factors reference · how RankGear measures · glossary.