TL;DR
This paper introduces a novel text input front-end processor (FEP) that acts as an information access platform, seamlessly integrating dictionary and example sentence lookups into the writing process. By automatically extracting keywords and intelligently retrieving relevant information, it significantly reduces user interruption and enhances writing efficiency, particularly for Japanese speakers writing in English.
What Problem Does This Paper Address?
Writing often involves consulting external resources like dictionaries and example sentences. Traditional methods, however, require users to actively switch applications, manually input queries, and copy-paste information, leading to significant interruptions that disrupt the flow of thought and reduce writing productivity. This paper addresses the challenge of creating a writing support tool that minimizes these interruptions by making information access an integral and largely automatic part of the text input process.
Methods
The core innovation is a Front-End Processor (FEP)-type Information Access Platform. This FEP acts as an intermediary between the user's keyboard input and any text-input application.
- Morphological Analysis: As the user types, a morphological analyzer automatically identifies and extracts keywords from the input text.
- Information Retrieval: These keywords are then used to query various information sources, including system dictionaries, CD-ROM dictionaries, and a bilingual example sentence database.
- Controllable Automation: A key design feature is the ability to control the extent of automatic retrieval and display. Users can choose from three modes:
- Fully Automatic: Information is retrieved and displayed without user command.
- Semi-Automatic: Information is retrieved automatically but displayed only upon user command.
-
Manual: Both retrieval and display require explicit user commands. This tiered approach optimizes system efficiency by adapting to the type of information being accessed.
-
"Eisaku Pen" (English Writing FEP): This component specifically handles Japanese-to-English conversion, initially mapping Japanese words to English equivalents while preserving Japanese grammatical structures. It then applies a stepped-level interactive machine translation to transform the sentence into natural English.
- "Shoseki Renzu" (CD-ROM Dictionary Tool): This module provides access to CD-ROM dictionaries, using keywords extracted by the FEP.
- "Reibun Bainda" (Bilingual Example Sentence Database): This component retrieves relevant Japanese-to-English example sentences based on input keywords.
Datasets
- System Dictionary: Contains approximately 100,000 Japanese vocabulary entries and 15,000 idiomatic expressions. The idiom dictionary was reportedly built manually.
- "Reibun Bainda" (Bilingual Example Sentence Database): Contains 3,000 Japanese-to-English bilingual sentence pairs specifically for letter writing.
Key Insights and Ideas
- Seamless Integration is Key: The most significant insight is that integrating information access directly into the text input stream, rather than treating it as a separate task, dramatically improves writing efficiency.
- Controllable Automation Balances Efficiency and Control: The ability to adjust the level of automation is crucial. While full automation is convenient, it can be overwhelming. Offering user control prevents information overload and ensures the tool remains helpful without being intrusive.
- Keyword Extraction as a Universal Access Point: Automatically extracting keywords from input text provides a robust and general mechanism for accessing diverse information sources, reducing the burden on the user to formulate search queries.
- FEP as a Versatile Platform: The FEP architecture is designed to be application-agnostic, allowing it to be combined with various text-input applications and other FEPs (like Kana-Kanji converters), making it a flexible solution.
Key Results
The paper emphasizes qualitative improvements in user experience and efficiency rather than presenting extensive quantitative results. The primary outcome is a writing support tool that:
- Reduces Interruption: Users can concentrate better on writing due to less need for manual dictionary lookups or sentence retrieval.
- Enhances Efficiency: The automatic retrieval and display, coupled with user control, streamline the writing process.
- Facilitates Japanese-to-English Writing: The "Eibun Meibun Meikingu" tool, combining the FEP with machine translation, specifically aids Japanese speakers in producing English text.
Limitations and Future Work
- Limitations:
- The paper acknowledges that the similarity judgment for retrieving example sentences in "Reibun Bainda" uses a simple keyword-based method, which might lead to irrelevant results and potential interruptions if displayed automatically.
- The idiom dictionary was built manually, which is a labor-intensive process.
- Future Work:
- Expansion of the tool into a general writing support tool (beyond just English writing).
- Enlarging the resources the tool can access.
- Developing an example-based translation function that leverages "Reibun Bainda" for "Eisaku Pen's" Japanese-to-English conversion.
- Implementing an automatic example sentence acquisition function to add user input texts and their translations to "Reibun Bainda."
Summary updated