1 Introduction

This section introduces the potential applications of frame semantic resources across multiple domains, elaborates on the background of constructing the Chinese FrameNet, highlights the challenges in applying English semantic frames to the creation of Chinese semantic frames, reviews the development steps of English semantic frames and the current research status of Chinese semantic frame construction methods, emphasizes the necessity of exploring approaches to creating Chinese semantic frames, and outlines the research methods and foundations of this study.

2 Pre-construction: Necessity Verification

This section focuses on the necessity verification process before creating modern Chinese semantic frames. First, by combining bilingual dictionaries with the FrameNet lexical unit database, the target word "沏" (qiē) was examined. The study found that its English equivalents "infuse" and "brew" in bilingual dictionaries had no matching results in the FrameNet lexical unit database, confirming the necessity of creating a frame for "沏". Subsequently, the chapter elaborates on the importance of constructing a proto-lexical unit list, noting that the number of proto-lexical units is related to the necessity of frame creation. By consulting bilingual dictionaries and adhering to the inclusion principles of the Chinese FrameNet lexical unit database, "沏", "冲" (chōng), "泡" (pào), and "冲泡" (chōngpào) were identified as proto-lexical units, providing a foundation for the subsequent creation of semantic frames.

3 Construction Phase: Description of Frame Semantic Scenarios

This section delves into a crucial step in constructing modern Chinese semantic frames: the description of frame semantic scenarios. First, by analyzing the schema of the word "沏", it is determined that it is not monosemic but contains four senses, each corresponding to different semantic scenarios. The study focuses on the schema of "沏1", which is a composite schema involving temporal and spatial dimensions, describing the process of creating a liquid through three stages: placing entity A, pouring entity B, and forming entity C. Based on this schema, the frame semantic scenario of "沏1" is described as the process of an agent creating a liquid, involving the placement and pouring of the transferred entity, and the formation of a new entity after contact.

Next, the chapter discusses the determination of frame elements and lexical units. Frame elements are categorized into core and non-core elements, with core frame elements identified based on specific principles such as syntactic position, valence patterns, and necessity for word meaning comprehension. Through corpus analysis, core frame elements such as "Agent", "Old Transferred Entity", "Container", "New Transferred Entity", and "New Created Entity" were identified, and their definitions and semantic types were supplemented. The determination of lexical units is based on the ability of linguistic components to activate conceptual structures, taking into account the peculiarities of Chinese, such as the blurred boundary between words and non-words. Principles for marking lexical units were established by referencing authoritative dictionaries and combining corpus data, word segmentation tools, and native speaker intuition.

Finally, the chapter explores the determination of frame relations and frame names. Frame relations include nine types, such as inheritance and use, while the naming of frames follows principles of accuracy, convenience, and conciseness. By analyzing the semantic scenario and schema of "沏1", it was determined that the semantic frame to which "沏1" belongs inherits from "Liquid Making" and "Intentional Creation", and has part-whole relations with "Placing", "Causing Fluid to Move Downward", and "Soaking". Ultimately, the semantic frame for "沏1" was named "Brewing", completing the initial creation of this semantic frame.

4 Post-construction: Refinement, Summary, and Reflection

This section discusses the refinement, summary, and reflection on semantic frame creation. It points out that frame creation requires revisions and improvements as factors like corpus and experience evolve, using "Brewer" as an example to illustrate updates to frame elements. The innovative nature of semantic frame creation is emphasized, and a pathway for constructing Chinese semantic frames is summarized and refined, though its general applicability requires validation. Additionally, the necessity of manually creating semantic frames in the era of generative artificial intelligence and the role of generative AI in this task are explored. A comparison between the semantic frame for "沏" created by ChatGPT4 and the manually constructed version revealed differences in abstraction levels, though the AI model still shows potential for application.

5 Conclusion

This section summarizes the methodology for creating modern Chinese semantic frames, using the character "沏" as an example to construct the "Brewing" semantic frame through steps such as necessity verification, proto-lexical unit list construction, and semantic scenario description. Methods proposed include the verification approach combining bilingual dictionaries with the FrameNet lexical unit database, the method of constructing proto-lexical unit lists based on translation equivalents, and the schema analysis method for describing semantic scenarios. These represent an inheritance and localization attempt of frame semantics. Additionally, it is noted that combining human expertise with generative AI models to assist in semantic frame creation is efficient and feasible, serving as a future research direction. High-quality Chinese frame semantic resources will provide a solid foundation for related fields.