Sharp Kabushiki Kaisha Corporation

Taro Morishita

Shigeki Kuga

Masahiro Wada

Toshiyuki Tanaka

Nobuo Nakamura

Hiroyuki Kanza

Mikio Ohsaki

Satoshi Onishi

A text compression dictionary generation apparatus including: a frequency guideline calculation unit for calculating, with respect to a target compression rate, a frequency guideline for each character string having a length of at least two characters, required to occur in the text to cutout the target compression rate; a long character string cutout unit for cutting out a character string from the text having a length of at least two character which occurs at a frequency matching the corresponding frequency guideline of a character string, together with the actual frequency in the descending order of the length of the character string to generate a first static dictionary, while simultaneously generating a group of the remaining text not cut out and not matching the frequency guideline; a one-character symbol extraction unit for counting the frequency of each character of one-character string length from the group of remaining text not cutout to generate a second static dictionary; and an eventual static dictionary build up unit for generating an eventual static dictionary by adding the second static dictionary to the first static dictionary.

Text compression dictionary generation apparatus

An index structure of a dictionary retrieval apparatus includes a first index, a second index, a heading main index, and a differential character train buffer. An item for each word in the index is accessed via the first and second indexes. Access to succeeding and receiving items in the index is carried out at high speed since the item has data of a fixed length. The item includes the number of overlapping characters with a previous word, a differential character train length with respect to a previous word, and a pointer indicating a storage position of a differential character train with respect to a previous word in the differential character train buffer. The buffer has a differential character train with respect to a preceding word stored without duplication of any partial character train in the buffer. Therefore, the memory capacity used by the buffer can be suppressed.

Dictionary retrieval apparatus

Method and apparatus are provided for editing input information. Input information is received from an input means (1) from which a keyword is selected for editing purposes. A low level record is located in a memory (5) which has the keyword of the input information stored therein. The low level record having the keyword stored therein is accessed to determine the location in the memory of higher level records and information types of the higher level records. The higher level records having stored therein explanatory information for the keyword in accordance with the information types of the higher level records. The higher level records are of a plurality of information types for presenting different types of explanatory information regarding the keyword, such as a definition of the keyword, a synonym for the keyword, an example usage of the keyword, etc. A display (55) is provided with an indication of the information types stored in the memory for the keyword so that a user can select a desired information type. The explanatory information associated with the desired information type is then displayed to the user, thereby facilitating the user's decision whether the keyword is properly used in the input information.

Document processing support system using keywords to retrieve

A text processing system for example a word processor includes a storage unit with two storage devices. The first storage device can store at least two character strings in an arbitrary order. The second storage device can store at least one index. Each stored index designates a predetermined one of the character strings. The indexes stored are in a predetermined order.

Text editor with memory for eliminating duplicate sentences

An element checking system for use in forward inference system is disclosed. When checking elements, each element is read out as a token and transmitted through a checking network. The element checking system includes memory for storing an element checking network. The network has a plurality of inputs for receiving the tokens one at a time, and a plurality of nodes for checking features of the received token. The system also includes working memory for storing elements, each element carrying data identifying a particular node from which the checking procedure in the networks starts. An additional memory is included which stores a pattern table including a plurality of different patterns with a pattern number assigned to each pattern, and a linking table including a pattern number and a corresponding node number for designating a starting node number for an element having a corresponding one of the plurality of patterns. Through the use of the pattern and linking table, an efficient element checking system utilizing a minimum number of nodes can be achieved.

Element checking system for use in forward inference system

A computer, in particular a word processor, is structured to enter and store Japanese symbols and character strings. The word processor includes an operator defined dictionary. To enter the new words in particular Japanese words into the operator defined dictionary, reference familiar to the operator are displayed. The reference words which are similar in linguistic form to the new words, have parts of speech which are prestored in a dictionary of the word processor. The operator obtains the parts of speech of the reference words which are comparable to the new words. Thereafter, the new words with identified parts of speech in proper form are entered into the operator defined dictionary.

System for registering new words by using linguistically comparable

A conversation type inquiry-and-answering system that can be used, for example, in a diagnosing system. The inquiry-and-answering system can process data in a conversational manner between an operator and the system on the basis of a knowledge data base. The knowledge data base includes a first memory for storing information, such as facts and rules, and a second memory for storing the contents of inquiry and their responses. Also, there is an editing system for the data base.

Inputting system and an editing system in an inquiry-and-answer system

An expert system having a forward inference control method in which an inference process (hereinafter referred to as trace) is displayed in consideration of the condition of a negation when the trace is displayed to verify a rule inputted to the expert system, whereby, a user can confirm that a rule is unexecutable because of the condition of the negation being not satisfied.

Display method in an inference trace of an expert system

The present invention discloses an inputting system for an inquiry-and-answer which removes the necessiates for strict analysis of the meaning of a response to an inquiry in order to extract necessary facts for inferences. By considering the response by an operator along with its corresponding inquiry from the system, the inputting system is enabled to editing the knowledge data base in a much easier fashion than the prior art. The inputting and editing system analyzes the answer to an interrogative and matches the pronouns and other elements of the answer with their corresponding elements from the interrogative so that the operator does not have to answer the interrogative with the exact language of the interrogative for the system to understand and utilize the answer properly.

Growing community of inventors

Nara, Japan

Taro Morishita