AI PatentTrans. Frequently Asked Questions

A: In patent translation, it is crucial to use the same translated term for the same concept throughout the document. Since AI tends to pull from various past examples, leading to terminology fluctuations, we strictly apply Translation Memories (TM) and Glossaries. Furthermore, translation proofreaders and QA managers verify overall consistency, with a particular focus on the correspondence between the Claims and the Specification.

A: AI tends to paraphrase based on context, which poses a risk of altering the technical meaning. To counter this, translators well-versed in the specific technical field verify the original technical intent. We prioritize legal accuracy regarding claim interpretation over grammatical "naturalness" during the correction process.

A: We detect structural inconsistencies at the initial drafting stage and propose restructuring during the translation proofreader’s review. Since there are specific practical requirements—such as the number of independent claims and multiple dependent claim formats for the U.S., or the "one independent claim per category" rule for the EPO—we identify the need for structural adjustments simultaneously with the translation.

A: We consider how the structure and wording of the specification affect claim interpretation. We make adjustments aligned with U.S. practice risk mitigation, such as organizing descriptions of "Problems" and "Means" and adding Incorporation by Reference where appropriate.
For details, please see the service details.

A: Examples include converting claims into the Two-Part format, adding reference signs, and inserting general disclosure sections to secure room for future amendments. We also propose consolidating the claims to 15 or fewer to help control filing costs. By aligning the structure with EPO requirements, we aim to reduce additional costs and amendment risks during examination.
For details, please see the service details.

A: We reflect past projects and client-specific glossaries to the maximum extent possible to maintain uniformity in terminology and style.

A: By using AI to generate the first draft quickly, our expert translators can focus their energy on critical technical sections. Our multi-layered QA process prevents any decline in quality that might otherwise result from a shortened schedule.

A: Since Office Action documents directly impact the outcome of the application, they require accuracy equal to or greater than the patent specification. We emphasize the logical flow of the response and the accuracy of technical expressions, increasing review intensity if necessary.

A: Translation data is never used for AI training. We ensure that confidential, pre-publication information is never leaked externally.

A: If an examiner does not check the original text, a mistranslation may only become a problem later (e.g., during litigation). Therefore, we strictly verify terminology consistency, logical coherence, and technical accuracy at the initial stage to minimize the risk of future legal defects.

A: During translation, we detect expressions that could lead to fixed or limited interpretations. We select terminology with future examination and enforcement in mind to avoid unintentionally narrow constructions.

A: The process follows this sequence: Translation Proofreader Review → Quality Assurance → (Optional) Patent Attorney Review. Each stage has a different verification perspective, which results in a higher error detection rate than a single-step process.

A: We require the document type, language pair, requested deadline, word/character count, and any special instructions (e.g., availability of glossaries or past related cases). This allows us to provide a quote that optimizes the balance between quality and lead time.

A: Our baseline is the Standard Patent Plan (17 JPY/word), but this can be adjusted with additional quality options, such as a Patent Attorney review. We propose the most suitable configuration based on the risk and importance of the case.