In Ex parte Butler, the application was directed to a positioning control system for lithographic printing, and the claim term at issue was "a partial order filter". The claim was rejected as anticipated by a reference (Yuan) teaching a photolithography system. The reference system used a filter and listed notch, low pass, and high pass filters as examples.
The Applicant argued that Yuan did not disclose a filter of the claimed type ("partial order"). In the Answer, the Examiner provided a Wikipedia definition of a mathematical filter as "a special subset of a partially ordered set." The Examiner then reasoned, based on the definition, that "in the broadest sense, the filter of Yuan can be considered as a 'partial oder filter' as broadly claimed." As additional support for his interpretation, the Examiner explained "Yuan’s filter is a partial order filter because, like the Appellants’ filter [as described in the specification), it can be a low pass filter."
The Board found the Examiner's interpretation to be unreasonable. Addressing the Examiner's definition, the Board explained that "the Examiner has not established that those of ordinary skill in the photolithography art to which Yuan pertains would have considered a partial order filter to be defined by the definition of a mathematical filter." (Emphasis added.) Addressing the Examiner's comment on a statement in the Applicant's specification, the Board referred to the teachings of the specification a whole:
The specification "does not indicate that every low pass filter is a partial order filter but, rather, indicates that a partial order filter differs from a first order filter in that a first order filter provides a -90º phase shift whereas a partial order filter provides a lesser phase shift (-60º for a 2/3order filter and -45º for a half order filter)The application in Ex parte Xu was directed to network routing, and the claim terms at issue were "hashing" and "hash function". The Examiner asserted that these terms read on mapping as described in Lu: "The current geographic position of the node is converted to a zone of the network based on a mapping function or table.” As support for his assertion, the Examiner provided a definition of "hash" from a technical dictionary:
hash2 vb. To be mapped to a numerical value by a transformation known as a hashing function. Hashing is used to convert an identifier or key, meaningful to a user, into a value for the location of the corresponding data in a structure, such as a table.On appeal, the Applicant focused on the phrase “by a transformation known as a hashing function” in the definition, and reasoned that the “mere mapping of data in Lu” could not be characterized as “hashing”. In the Answer, the Examiner maintained the rejection and provided this additional explanation:
[T]o hash is to be mapped to a numerical value by a formula (a transformation known as a hashing function). . . .When a value is mapped to another value, the two values are made equal to each other. For example, f(x)=x [is a mapping]. The formula pt=h(pt,dp,dz,z) [in Applicant's specification] is of a form similar to f(x)=x, but involves multiple variables and has no real disclosure of what h(pt,dp,dz,z) is ... As x=y and f(x)=x (which are both functions that simply map the value of x to either y or f(x), respectively) are both both formulas they meet the definition of hashing functionThe Applicant countered by arguing, in the Reply Brief, that the Examiner had improperly interpreted the phrase “[t]o be mapped to a numerical value by a transformation known as a hashing function” (in the definition) to read hashing as the use of any mathematical function to map to a numerical value.
(Examiner's Answer, emphasis added.)
The Board agreed with the Applicant:
[T]he definition defines hashing as a particular way of mapping to a numerical value, namely, as using a transformation known as a hashing function to map to a numerical value. Because Lu does not describe the mapping function as a hash function or provide any details suggesting that the mapping process employs a hash function, we agree with Appellants that the Examiner has failed to demonstrate anticipation of the “hashing” claims.
My two cents: I view these two cases as ones where the Examiner decided that Broadest Reasonable Interpretation allowed him to broaden from a dictionary definition, and the Board said No. It's not clear whether the Board was influenced by teachings in Applicant's spec. The Board certainly didn't say it was, nor did it cite to any case law describing use of the spec to interpret claims. However, perhaps the spec did help these Applicants: Butler's spec did explain what was meant by "partial order filter" and Wu's spec did include an example of a specific hashing function.