<?xml version="1.0" standalone="yes"?>
<Paper uid="C80-1071">
  <Title>SPEECH RECOGNITION SYSTEM FOR SPOKEN JAPANESE SENTENCES</Title>
  <Section position="4" start_page="0" end_page="0" type="metho">
    <SectionTitle>
OZIISAN OBSIISAN OISA
YAMA IEAMA AA
SENTAKU SEN.PA.PU SE.A.U
OOKII OO.PSI O.SI
</SectionTitle>
    <Paragraph position="0"/>
    <Section position="1" start_page="0" end_page="0" type="sub_section">
      <SectionTitle>
3.3. Semantic Knowledge
</SectionTitle>
      <Paragraph position="0"> Semantic information is used for the following purposes.</Paragraph>
      <Paragraph position="1"> (i) Elimination of semantically inconsistent sentences which have been recognized using only acoustic and syntactic information.</Paragraph>
      <Paragraph position="2"> (ii) Future development to semantic understanding of natural language by forming semantic networks. null (iii) Control of transition on the syntactic state transition network through the syntax analyser. null  One of the semantic information dealt with is &amp;quot;knowledge about meaning&amp;quot;. This knowledge involves (i) what each word means, (ii) verb-centered semantic structure, and (iii) schema of a story \[i0\]. The other information is, so called, &amp;quot;remembrance of episode&amp;quot; which means the remembrance of a topic of conversation. In the present system, meaning of a word is represented by a list structure, and the others are represented by networks.</Paragraph>
      <Paragraph position="3"> In the system the knowledge about meaning must be given from outside and can not yet be increased or updated by itself, but remembrance of episode can be increased or updated whenever new information comes in. While, if a schema has been already formed for a topic to be talked from now on, the knowledge of the topic will help recognition of the spoken topic. In the following sections how semantic information works in the recognition system will be explained.</Paragraph>
      <Paragraph position="4"> 3.3.1.1. Meaning of a word Denote a word by n, its characteristic features by fi(i=l,...,m; m is the number of features). Then, the meaning of a word may be expressed as follows: n(fl' f2' &amp;quot;''' fm )' where f. = 1 when the word has the characteristic 1 feature f, l f = 0 when the word has not the feature f . 1 1 For example, if fl = concrete, f2 = creature, f3 = animal, .... then hill (1, 0, 0, ..... ), dog (i, i, i, ..... ).  3.3.1.2. Definition of a verb A verb plays very important semantic role in a simple sentence. A semantic representation of meaning of a verb is shown in Fig. 3.7, where n O , n I , ..., n. are nodes, and Ar I, Ar 2, .., Ar. l 1 attatched to each arc are the natures of each arc. The nature of a node n is determined by a P nature Ar attatched to the arc directing to the  n. a word or node qualified by l a nature Ar. 1.</Paragraph>
      <Paragraph position="5"> For example, a verb &amp;quot;IKU (go)&amp;quot; is defined by Fig. 3.8.</Paragraph>
      <Paragraph position="6"> 3.3.1.3. Schema The form of a schema can not be determined uniquely. Dealing with a story, we may be able to represent the schema, for example, as shown in Table 3.4 and Table 3.5.</Paragraph>
      <Paragraph position="7"> 3.3.1.4. Remembrance of an episode --- Formation of a semantic network Refering to the results of syntactic analysis and the relation between the nature of an arc and a case particle (partly involving another particle), the system forms a semantic network for a simple sentence centering a recognized verb. For instance, if a word sequence OZIISAN WA YAMA E SHIBAKARI NI IKIMASHITA.</Paragraph>
      <Paragraph position="8"> (An old man went to a hill for gathering) firewoods.</Paragraph>
      <Paragraph position="9"> with syntactic information is given, a network shown in Fig. 3.9 will be formed. In Fig. 3.9 a process constructing a sentence is also shown. 3.3.2. Linking a semantic network for a sentence with a semantic network for an episode After a network for a sentence has been formed, the network must be linked Up with the already constructed network for the current episode. For this purpose a new node must be identified with the same node in the episode network.  and its semantic network (b) for &amp;quot;An old man went to a hill for gathering firewoods.&amp;quot;. --- shows a phrase, shows modification and RENYO in \[ \] means this phrase modifies an inflexional word or phrase, ino: in order to. In the present system all relations explicitly appearing in sentences and nodes expressing location are examined whether they have already appeared or not. Time relation is not handled unless it appears explicitly in sentences. Deeper structures of meaning such as causality or reasoning are not yet able to be dealt with. Fig. 3. i0 illustrates a network for the episode, which has been constructed after the system has processed several sentences at the beginning of the tale of &amp;quot;MOMOTARO&amp;quot; shown below.</Paragraph>
      <Paragraph position="10"> There lived an old man and an old woman.</Paragraph>
      <Paragraph position="11"> The old man went to a hill for gathering firewoods. null The old woman went to a brook for washing.</Paragraph>
      <Paragraph position="12"> She was washing on a brookside.</Paragraph>
      <Paragraph position="13"> 3.3.3. Word prediction by a conjunction &amp;quot;TO (and)&amp;quot; When the syntax analyser has found a conjunction &amp;quot;TO (and)&amp;quot; which is used to enumerate some nouns, the system can predict a following noun group. For instance, for the input &amp;quot;MOMOTA-</Paragraph>
    </Section>
  </Section>
  <Section position="5" start_page="0" end_page="1033" type="metho">
    <SectionTitle>
RO WA INU TO ... (MOMOTARO was accompanied by a
</SectionTitle>
    <Paragraph position="0"> dog and ... &amp;quot;, the system picks up as a following noun a noun group having similar natures to those a dog has.</Paragraph>
    <Paragraph position="1"> 3.3.4. Application of semantic knowledge to speech recognition Using semantic knowledge the system advances recognition process as follows: (i) Using acoustic and syntactic information, and sometimes semantic information, the system processes an input sentence and outputs several word sequences. The syntax analyser gives to each word sequence necessary syntactic information such as part of speech of each component word, phrase and modifying relation between an old man  tic information, forms a semantic network for each word sequence.</Paragraph>
    <Paragraph position="2"> (iii) A word sequence for which a semantic network failed to be formed satisfactorily is rejected because of semantic inconsistency. For instance, for an input sentence: &amp;quot;OZIISAN WA YAMA E SHIBAKARI NI IKIMASHITA.(An old man went to a hill for gathering firewo6ds.)&amp;quot;, an output word sequence: &amp;quot;OZIISAN WA HANA (flower) E SHIBAKARI NI IKIMASHITA.&amp;quot; is rejected, because the verb &amp;quot;IKU (go)&amp;quot; has an arc &amp;quot;to Location&amp;quot; but the output word sequence has no word meaning location and also the word &amp;quot;HANA (flower)&amp;quot; has no appropriate arc in the network.</Paragraph>
    <Paragraph position="3"> (iv) Taking into account the result of syntax analysis and reliability of acoustic matching, the most reliable word sequence is output.</Paragraph>
    <Paragraph position="4"> (v) Finally, the semantic network of the output sentence is linked with the semantic network of the episode formed by this process stage.</Paragraph>
  </Section>
class="xml-element"></Paper>