Smarter Way To Translate SEM Keywords By NGram Analyzer

Reading this article, it can help you

1. Develop a complete set of keywords as they are badly structured in the running account
2. Refine the keyword list from any useful material in the account
3. Translate a huge amount of keyword to other language in a smart way because you feel not secured to pass them to Google Translate directly

Check My Real Case Example:

I have a huge amount of keywords with

• wrong grouping(1), 
• some keywords missing(2) 
• and the worst thing is I have to replicate them into English(3).

I guess most of us came across this.We develop our own system to tackle the problem as a learning path which eventually developed our style.

STEP 1 - Understanding The Keywords

Keywords are prefix + core term + suffix.
When we need to expand keyword list, keyword multiplier is the first thing comes to our head.
Even if we compile a keyword list creatively, the final outcome is similarly structured in this way.

Thus the first thing we do is reverse engineering - break down the existing keywords in prefixes, core terms, suffixes.

NGram Analyzer - The Beginning of Data Mining

The concept of NGram Analyzer would be a perfect a diagnostic tool to the keyword structure.

This is my existing keyword list:
TV 比較
電視 價格
高清電視 價格
電視機 價格
高清電視機 推薦
高清電視 推薦
TV 尺寸
3D TV 價格
3D TV 比較
3D TV 尺寸
電視 尺寸
TV 推薦
電視 比較
顯示屏 比較
HDTV 比較

STEP 2 - Put Them Into NGram Analyzer

Try other options and explore. Depends on your case, unigrams and occurrence of 1 are not the only choice, repeat this step to obtain a higher completeness of result.

And I got this break down.

Translate them into English in this step.

You may notice that some translations are not appropriate at all.













compare is correct as prefix and should be comparison as suffix, display is monitor, recommend is recommendation.

STEP 3 - Reorganize The Prefixes, Core Terms, Suffixes Into Right Place For Keyword Multiplier

Prefix	Core Term	Suffix
3D	TV	price
compare	HD TV	size
	monitor	recommendation
	TV Set

Now the list has no missing combination and been translated systematically.