The district court erroneously substituted its own judgment as to the “ordinary meaning” of the “ranking” terms contained in the ‘338 patent, instead of giving those terms their proper meaning in light of the intrinsic record, viewed through the eyes of one of ordinary skill in the art. Had the district court viewed the claims in this way, it would have arrived at the proper construction of the “ranking” terms, as meaning categorizing a method from an object-oriented computing environment based on its determined suitability.

In the context of the ‘338 patent, “ranking” is used much more like the “rankings” given in academic settings, not (as the district court thought) the narrow and specific kind of ordinal rankings, e.g., of sports teams in “top 25 rankings.” A teacher might rank students by giving a numerical ranking (100, 90, 80, etc.) to each one, by giving a letter-grade ranking (A, B, C, D, F), or by simply assigning the student to one of two categories, as in the case of assigning pass/fail grades (P, F). In this context, a single student may be ranked, and multiple students may have the same ranking. Viewed through the eyes of one of skill in the art, the ordinary meaning of “ranking” used in the ‘338 patent is “to assign to a particular class.”

The specification's teachings confirm this ordinary understanding of “ranking.” When a user of an array-based mathematical tool attempts to invoke a method in an object-oriented environment, the tool retrieves a list of method signatures. That list may represent zero, one, or multiple methods that have the same name as the desired method. According to the invention, the mathematical tool will compare the data types of the possible methods with the input parameters of the mathematical tool. Based on the results of that comparison, the mathematical tool will categorize, or assign, the methods to a class based upon suitability. This is “ranking.”

In some embodiments, the mathematical tool will calculate a narrower and more specific type of ranking, referred to in certain dependent claims as a “fitness ranking.” Fitness ranking is a species of “ranking” referring to the assignment to a class that corresponds to a particular level of suitability. There is nothing, however, in the specification or file history that limits the broader term, “ranking” to that specific assignment technique. Yet that is precisely what the district court's claim construction did.

The district court's construction is wrong for several reasons.

First, the district court determined the “ordinary” meaning of “rank” based on nothing other than its own understanding of the meaning of that term, and it provided no support other than its own assertion for that “ordinary” understanding.

Second, the district court's claim construction violates the doctrine of claim differentiation. By construing the “ranking” terms to require that the method signatures be ordered relative to one another, the district court vitiated the distinction between “ranking” and “fitness ranking.” “Ranking” requires only a categorization based upon suitability vel non, while “fitness ranking” requires categorizations of specific levels of suitability, which is the same as what the district court required out of the broader term “ranking” – putting the method signatures in order relative to one another.

Third, the district court's claim construction is erroneous because it excludes the preferred embodiment of the ‘338 patent – a kind of construction that this Court has said is “rarely, if ever, correct.” The district court's construction excludes the preferred embodiment because, inter alia, it does not allow for the case when zero or one method signatures are retrieved, or when multiple method signatures have the same rank.

Fourth, the district court wrongly determined that the step of “comparing the data types … to determine suitability of each method” precluded a construction where the “ranking” step assigned methods to a suitable or unsuitable class. The district court failed to appreciate the difference between the former step, which involves the system making a mere determination of suitability, and the latter step of assigning the method signatures to a rank “based on the determined suitability.”

Finally, the district court's selective reliance upon two sentences out of the prosecution history was improper: Not only were those sentences perfectly consistent with the broader interpretation of “ranking,” but the district court's approach neglected the intrinsic record as a whole. That complete record confirms that, in the context of the ‘338 patent, “ranking” means simply sorting the methods based on suitability or the absence thereof; “ranking” means “to assign to a particular class.”

Math Works proffers a construction of the “ranking” terms in the 338 patent that is not consistent with the ordinary meaning of the terms, is contrary to the way in which those terms are used in the claims and in the specification, and is not supported by MathWorks' own use of those terms in its prosecution of the patent. MathWorks now wants to be its own lexicographer in defining the “ranking” terms, yet it failed to provide notice to the public of those special definitions in the‘338 patent.

In contrast, the district court's construction of the ordinary meaning of the “ranking” terms is entirely consistent with the intrinsic record, including the specification, the use of those terms in the claims themselves, and the prosecution history. It is MathWorks' proposed construction, not the district court's, that violates the doctrine of claim differentiation by not giving effect to all of the claimed terms in the patent. MathWorks' construction of “ranking” as assigning to a class of suitable or unsuitable fails to differentiate and renders superfluous the “comparing” element of the claim, which is the step “to determine suitability.” And the court's construction does not, as MathWorks asserts, exclude the preferred embodiment.

Leave a Reply

Name (required)

Email Address(required)

Website

Comments