Patent Metrics is the numerical parameter relate to the quality which can be extracted from the patent specification.
PatentValueAnalyst can be evaluated the quality of the patent specification by using about a hundred kinds of Patent metrics.

• Patent metrics that higher(bigger) is better.
• The number of the independent claim
• The number of the category
• The nest level (The depth of the claim tree)
• The number of cited patent document
• The number of cited non-patent document
etc.
• Patent metrics that lower(smaller) is better.
• The number of ambiquous words
• The number of purposeful words
• The number of sentence longer than the threshold
• The number of paragraph longer than the threshold
• The number of sentence without subject
etc.

Please click here to view a sample of extracting Patent Metrics.

By calculating and substituting the extracted patent metrics into the formula for computation
that is configured each quality characteristic in the patent specificaton, the quality of the patent specification can be evaluated.
Please click here here to view a sample of Evaluation Fomula.

A detailed contribution ratio(coefficient) of the individual patent metrics for the patent quality evaluation can be configured.
In fact, this is effective if the definition of the patent specification quality and the evaluation method of the patent specification quality are determined.
In addition, this can be taken advantage of checking the quality of the patent specification before the process of a filing by customizing the output form.

By analyzing and studying the evaluation score for the patent specification(the patent publication) as to be the tutorial data and the patent metrics extracted from those texts,
the formula that describes the relationship between the evaluation score and the patent metrics can be figured. (This is what is called the regression analysis.)
The quality of the patent specification can be evaluated based on the result.

The positive results similar to the evaluation by a person (the tutorial data) can be received. (the correlation coefficient:0.94)
In addition, by reason of taking advantage of the past evaluation quality results, the cost of adoption can be reduced.
Also, by adopting the SVM, the highly accurate evaluation can be received more objectively.

The SVM is the one of the supervised learning algorithm.
The algorithm described above is the most excellent in many other currently-known supervised learning algorithms.
The PatentValueAnalyst-SVM is computed the regression analysis(the SVM regression) based on this SVM principle.

The Invention Feature Extraction Level, the Inventive Expansion Level, the Toughness and the Specification Disclosure Level can be evaluated in the default.
The patent metrics for use in the patent specification quality evaluation is able to customize.

We use the PatentValueAnalyst-SVM. We use the evaluation result of the person in charge of checking the specification quality as the learning data. All our members of the department are happy with the correlated high performance in accordance with the evaluation result of the person in charge and the relief of burden on the person in charge. By Electronics manufacturers

We make sure to check the specification quality with the software before sending a draft to our client and the process of application. We are able to deliver and apply a certain level of the specification quality with the assistance of the software. We get satisfaction from our clients. By Patent office

OS	・Windows 7（32bit、64bit） ・Windows Vista（32bit、64bit） ・Windows XP
Hardware specs	What OS and applications of the above will work.

　* \498,000 (tax excluded)

Please make application by application form.