T‡¯Ô˜ 6 ñ M¿ÚÙÈÔ˜ 2000 / Issue No6 ñ March 2000
™ÙÔ ÂÍÒÊ˘ÏÏÔ ÂÈÎÔÓ›˙ÂÙ·È ÂÓ›ÁÚ·ÊË Ï¿Î· Ë ÔÔ›· ‚Ú¤ıËΠÛÙËÓ ∞ÎÚfiÔÏË ∞ıËÓÒÓ Î·È ÂÚÈÏ·Ì‚¿ÓÂÈ ÂÚÌËÓ›· ÛÙÂÓÔÁÚ·ÊÈÎÔ‡ Û˘ÛÙ‹Ì·ÙÔ˜ Ô˘ ·Ó¿ÁÂÙ·È ÛÙÔÓ ¢’ .Ã. ·ÈÒÓ·. ∏ Û˘ÌÏ‹ÚˆÛ‹ Ù˘ ¤ÁÈÓ ·fi ÙÔÓ Gomperz Î·È ¤¯ÂÈ ˆ˜ ÂÍ‹˜:
The cover depicts an inscribed stone which was found in the Athens Acropolis and contains the interpretation of a stenographic system dated in the 4th century B.C. Its completion was done by Gomperz and is as follows:
∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 1 / page 1 M¿ÚÙÈÔ˜ 2000 / March 2000
"§O°O¶§OH°H™H" ¶›Ó·Î·˜ ¶ÂÚȯÔÌ¤ÓˆÓ / Table of Contents M¿ÚÙÈÔ˜ 2000 T‡¯Ô˜ 6
EÈÛÙËÌÔÓÈÎfi˜ Y‡ı˘ÓÔ˜: ∂ÈÛ·ÁˆÁÈÎfi ™ËÌ›ˆÌ· / Introductory Note ÛÂÏ./pg. 2 K·ıËÁËÙ‹˜ °ÈÒÚÁÔ˜ K·Ú·ÁÈ¿ÓÓ˘ ¶PO°PAMMA “¢IA§O°O™” (E¶ET II) ÛÂÏ./pg. 3 Y‡ı˘ÓÔÈ ŒÎ‰ÔÛ˘: ºÈÏÔÛÔÊ›· ÙÔ˘ “¢IA§O°OY” ÛÂÏ./pg. 3 ¢Ú. Iˆ¿ÓÓ· M·Ï·Á·Ú‰‹ MÂÙ¿ÊÚ·ÛË Î·È ÌÂÙ·ÊÚ·ÛÙÈο ÂÚÁ·Ï›· ÛÂÏ./pg. 3 ™˘ÓÂÚÁ¿Ù˜: HÏÂÎÙÚÔÓÈ΋ ÏÂÍÈÎÔÁÚ·Ê›· ÛÂÏ./pg. 3 °È¿ÓÓ˘ ¢ÔÏfiÁÏÔ˘ ¢ÈfiÚıˆÛË Ï·ıÒÓ ÛÂÏ./pg. 4 B¿Ûˆ ¶·Ó·ÁÔÔ‡ÏÔ˘ ¢È‰·Ûηϛ· Ù˘ EÏÏËÓÈ΋˜ ˆ˜ ͤÓ˘ ÁÏÒÛÛ·˜ ÛÂÏ./pg. 4 AÓ·ÛÙ¿ÛÈÔ˜ ¶·ÙÚÈοÎÔ˜ ™˘ÓıÂÙÈ΋ ʈӋ Î·È Û˘ÛÙ‹Ì·Ù· ÁÈ· Ù˘ÊÏÔ‡˜ ÛÂÏ./pg. 4 °Ú·Ê›ÛÙ·˜: TËÏÂÙÚ·Â˙ÈÎfi Û‡ÛÙËÌ· ʈÓËÙÈ΋˜ ÂÈÎÔÈÓˆÓ›·˜ ÛÂÏ./pg. 4 ÕÚÙÂÌȘ °Ï¿ÚÔ˘ ™‡ÓıÂÛË ÎÂÈ̤ÓÔ˘ ÛÂÏ./pg. 4 MÂıÔ‰ÔÏÔÁ›· ÙÔ˘ “¢IA§O°OY” ÛÂÏ./pg. 5 ¢È‡ı˘ÓÛË: IÓÛÙÈÙÔ‡ÙÔ EÂÍÂÚÁ·Û›·˜ ÙÔ˘ §fiÁÔ˘ ¶ÚÔ˚fiÓÙ· Ô˘ ·Ó·Ù‡¯ıËÎ·Ó ÛÂÏ./pg. 5 AÚÙ¤ÌȉԘ 6 & Eȉ·‡ÚÔ˘ AÍÈÔÔ›ËÛË ÙˆÓ ·ÔÙÂÏÂÛÌ¿ÙˆÓ ÙÔ˘ “¢IA§O°OY” ÛÂÏ./pg. 5 151 25 ¶·Ú¿‰ÂÈÛÔ˜ AÌ·ÚÔ˘Û›Ô˘ ™˘ÓÂÚÁ·˙fiÌÂÓÔÈ ÊÔÚ›˜ Ù˘ ÎÔÈÓÔÚ·Í›·˜ ÛÂÏ./pg. 5 ÙËÏ.: 6875300 ñ fax: 6854270 e-mail: [email protected] ¶PO°PAMMA “™A¶ºø” (E¶ET II) ÛÂÏ./pg. 6 http:// www.ilsp.gr “™A¶ºø” / M¤ÚÔ˜ Ô˘ ·ÊÔÚ¿ Ù· NH™IA ÛÂÏ./pg. 6 EÏÏËÓÈο & •ÂÓfiÁψÛÛ· DVD-ROM ÛÂÏ./pg. 6 TËÓ Â˘ı‡ÓË ÙˆÓ ÎÂÈÌ¤ÓˆÓ ¤¯Ô˘Ó ÔÈ “™A¶ºø” (K·ÈÓÔÙƠ̂˜/¶ÚˆÙÔÙ˘›Â˜/¶ÚˆÙÔÙ˘›Â˜ ˆ˜ ÚÔ˜ ÙÔ ˘ÏÈÎfi) ÛÂÏ./pg. 6 Û˘ÁÁÚ·Ê›˜. AÍÈÔÔ›ËÛË ÛÂÏ./pg. 6 H ¯ÚËÌ·ÙÔ‰fiÙËÛË Ù˘ ¤Î‰ÔÛ˘ ·˘Ù‹˜ “™A¶ºø” / M¤ÚÔ˜ Ô˘ ·ÊÔÚ¿ Ù· EÏÏËÓÈο ¯ÚfiÓÈ· (“K§EO¶ATPA”) ÛÂÏ./pg. 7 ¤ÁÈÓ ·fi ÙÔ ¤ÚÁÔ HOPE (IST) ÙÔ ÔÔ›Ô ¯ÚËÌ·ÙÔ‰ÔÙ‹ıËΠ·fi ÙËÓ DG XIII Ù˘ From Granada to Athens - LREC 2000 ÛÂÏ./pg. 7 E˘Úˆ·˚΋˜ EÈÙÚÔ‹˜. Overview ÛÂÏ./pg. 7 Conference Aims ÛÂÏ./pg. 7 H “§ÔÁÔÏÔ‹ÁËÛË” ‰È·Ó¤ÌÂÙ·È ‰ˆÚ¿Ó. Conference Topics ÛÂÏ./pg. 8 Conference Programme Committee ÛÂÏ./pg. 9 Local Committee ÛÂÏ./pg. 9 Athens ÛÂÏ./pg. 9 Conference Venue - Zappeion Megaron ÛÂÏ./pg. 9 Scientific Papers ÛÂÏ./pg. 10 "LogoNavigation" From Stockholm to Chios Island - 8th ELSNET European March 2000 Issue NÔ 6 Summer School on Language and Speech Communication - TeSTIA 2000 ÛÂÏ./pg. 10 Scientific Director: Important Dates ÛÂÏ./pg. 11 Professor George Carayannis Programme Committee ÛÂÏ./pg. 12 Edition Executives: Local Organisation Committee ÛÂÏ./pg. 12 Dr. Ioanna Malagardi Sponsors ÛÂÏ./pg. 12 More Information ÛÂÏ./pg. 12 Collaborators: Ioannis Dologlou ¶ÚÔÁÚ¿ÌÌ·Ù· MÂÙ·Ù˘¯È·ÎÒÓ ™Ô˘‰ÒÓ (E¶EAEK) Vaso Panagopoulou Anastasios Patrikakos “TEXNO°§ø™™IA” ÛÂÏ./pg. 12 ¶ÚfiÁÚ·ÌÌ· ™Ô˘‰ÒÓ ÛÂÏ./pg. 13 Graphics Designer: Artemis Glarou Text to Speech Synthesis in Greek ÛÂÏ./pg. 14 Address: Features ÛÂÏ./pg. 14 Institute for Language and Speech Processing Operation Enviroment ÛÂÏ./pg. 15 Artemidos 6 & Epidavrou Str. 151 25 Marousi Spectral Estimation Based on the Eigenanalysis of Athens, Greece Companion-Like Matrices ÛÂÏ./pg. 16 tel.: +301- 6875300 ñ fax: 301-6854270 1. Introduction ÛÂÏ./pg. 16 e-mail: [email protected] 2. Companion Matrix Based Spectral Estimation (CSE) ÛÂÏ./pg. 16 http://www.ilsp.gr 3. HTLS method ÛÂÏ./pg. 17 4. Experimental Results ÛÂÏ./pg. 17 The authors are responsible for text content. 5. Conclusion ÛÂÏ./pg. 18 Funding for this issue was carried out by the HOPE (IST) project which is funded Sappho Multimedia Bilingual Dictionaries for Tourists ÛÂÏ./pg. 20 by DG XIII of the European Commission. Macrostructure of the dictionary ÛÂÏ./pg. 20 "LogoNavigation" is distributed free of charge. Microstructure of the dictionary ÛÂÏ./pg. 20
∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 2 / page 2 M¿ÚÙÈÔ˜ 2000 / March 2000
∂ÈÛ·ÁˆÁÈÎfi ™ËÌ›ˆÌ· Introductory Note
∆Ô ¤ÎÙÔ Ù‡¯Ô˜ Ù˘ §ÔÁÔÏÔ‹ÁËÛ˘ ÂÚȤ¯ÂÈ ÙȘ The sixth issue of “LogoNavigation” contains the ÂÍ‹˜ ıÂÌ·ÙÈΤ˜ ÂÓfiÙËÙ˜: following thematic units:
∏ ÚÒÙË ıÂÌ·ÙÈ΋ ÂÓfiÙËÙ· ÂÚÈÏ·Ì‚¿ÓÂÈ ÙȘ ·- The first thematic unit contains the presentations ÚÔ˘ÛÈ¿ÛÂȘ, ÙˆÓ ŒÚÁˆÓ “¢π∞§√°√™” Î·È “™∞¶- of DIALOGOS and SAPPHO projects, carried out ºø”, Ô˘ ¤ÁÈÓ·Ó ÛÙÔ Ï·›ÛÈÔ Ù˘ HÌÂÚ›‰·˜ Ù˘ in the context of the Information Day under the °ÂÓÈ΋˜ °Ú·ÌÌ·Ù›·˜ ŒÚ¢ӷ˜ Î·È ∆¯ÓÔÏÔÁ›·˜ title “Society, Civilization and Cultural Industry Ì ı¤Ì·: “KÔÈÓˆÓ›· - ¶ÔÏÈÙÈÛÌfi˜ - ¶ÔÏÈÙÈÛÌÈ΋ organized by the General Secretariat of Research BÈÔÌ˯·Ó›·”. H ËÌÂÚ›‰· Ú·ÁÌ·ÙÔÔÈ‹ıËΠÙËÓ and Technology on March 3 2000 in the National 3Ë ª·ÚÙ›Ô˘ 2000 ÛÙÔ ∂ıÓÈÎfi ÿ‰Ú˘Ì· ∂Ú¢ÓÒÓ. Research Institution.
∏ ‰Â‡ÙÂÚË ıÂÌ·ÙÈ΋ ÂÓfiÙËÙ· ÂÚÈÏ·Ì‚¿ÓÂÈ ÁÂÓÈΤ˜ The second thematic unit contains general ÏËÚÔÊÔڛ˜ ÁÈ· ÙÔ ‰Â‡ÙÂÚÔ ‰ÈÂıÓ¤˜ Û˘Ó¤‰ÚÈÔ LREC information about the Second International “Second International Conference on Language Conference on Language Resources and Resources and Evaluation”, Ô˘ ı· Ú·ÁÌ·ÙÔÔÈËı› Evaluation “LREC”, which will take place in ÛÙËÓ ∞ı‹Ó· ·fi 31 ª·›Ô˘ ¤ˆ˜ 2 πÔ˘Ó›Ô˘ 2000. Athens from May 31 to June 2 2000.
∏ ÙÚ›ÙË ıÂÌ·ÙÈ΋ ÂÓfiÙËÙ· ÂÚÈÏ·Ì‚¿ÓÂÈ ÏËÚÔÊÔڛ˜ The third thematic unit contains information about ÁÈ· ÙÔ £ÂÚÈÓfi ™¯ÔÏÂ›Ô TeSTIA “European Summer the 8th ELSNET European Summer School on School on Language and Speech Communication”, Language and Speech Communication “TeSTIA”, Ô˘ ı· Ú·ÁÌ·ÙÔÔÈËı› ÛÙË Ã›Ô ·fi 15 ¤ˆ˜ 30 πÔ˘- which will take place on Chios Island from July 15 Ï›Ô˘ 2000 ˘fi ÙËÓ ·ÈÁ›‰· ÙÔ˘ ELSNET. to 30 2000.
∏ ٤ٷÚÙË ıÂÌ·ÙÈ΋ ÂÓfiÙËÙ· ÂÚÈÏ·Ì‚¿ÓÂÈ ÏË- The fourth thematic unit contains information ÚÔÊÔڛ˜ ÁÈ· ÙÔ MÂÙ·Ù˘¯È·Îfi ¶ÚfiÁÚ·ÌÌ· “∆∂- about the “TECHNOGLOSSIA” program of á√°§ø™™π∞”. postgraduate studies.
∆¤ÏÔ˜ Ë ¤ÌÙË ıÂÌ·ÙÈ΋ ÂÓfiÙËÙ· ÂÚȤ¯ÂÈ ∂Ú¢- Finally, the fifth thematic unit contains the ÓËÙÈο ·ÔÙÂϤÛÌ·Ù· Û¯ÂÙÈο Ì ı¤Ì·Ù· EÂÍÂÚ- research results related to the issues of Language Á·Û›·˜ °ÏÒÛÛ·˜, EÂÍÂÚÁ·Û›·˜ ºˆÓ‹˜ ηıÒ˜ and Speech Processing, as well as issues related Î·È ı¤Ì·Ù· º·ÛÌ·ÙÈ΋˜ AÓ¿Ï˘Û˘. to Spectral Analysis. ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 3 / page 3 M¿ÚÙÈÔ˜ 2000 / March 2000
¶ƒ√°ƒ∞ªª∞ “¢π∞§√°√™” (∂¶∂∆ ππ) ªÂÙ¿ÊÚ·ÛË Î·È ÌÂÙ·ÊÚ·ÛÙÈο ÂÚÁ·Ï›· ™ÙËÓ ÂÚÈÔ¯‹ ·˘Ù‹ ·Ó·Ù‡¯ıËÎ·Ó ÙÚÂȘ ‰È·ÊÔÚÂÙÈΤ˜ ¶·ÚÔ˘Û›·ÛË ÛÙËÓ ËÌÂÚ›‰· Ù˘ °°∂∆ ‰Ú¿ÛÂȘ Î·È Ï‹ıÔ˜ Û˘ÓÂÚÁ·ÛÈÒÓ. ™Ù· ı¤Ì·Ù· Ù˘ ‰È·- “KÔÈÓˆÓ›· - ¶ÔÏÈÙÈÛÌfi˜ - ¶ÔÏÈÙÈÛÌÈ΋ ÏÔÁÈ΋˜ ÌÂÙ¿ÊÚ·Û˘ ÚÔÛ‰ÈÔÚ›ÛıËΠ¤Ó· ÌÔÓÙ¤ÏÔ Û¯Â‰È·ÛÌÔ‡ ÂÓ’ fi„ÂÈ ÌÂÏÏÔÓÙÈ΋˜ ˘ÏÔÔ›ËÛ˘. ∏ ‰È·- BÈÔÌ˯·Ó›·” 3 ª·ÚÙ›Ô˘ 2000 ÏÔÁÈ΋ ÌÂÙ¿ÊÚ·ÛË ı· ‚ÚÂÈ ÔÏϤ˜ ÂÊ·ÚÌÔÁ¤˜ ‰ÈÂ- ∞fi ÙËÓ ÙÂÏÈ΋ ¤ÎıÂÛË ıÓÒ˜. ¢ÂÓ Â›Ó·È È‰È·›ÙÂÚ· ·Ó·Ù˘Á̤ÓË ÛÙËÓ ÂÔ¯‹ Ì·˜ ∫·ıËÁËÙ‹˜ °. ∫·Ú·ÁÈ¿ÓÓ˘, ¢È¢ı˘ÓÙ‹˜ π∂§ ÁÈ·Ù› ‰ÂÓ Î·Ù¿ÊÂÚ ӷ ‚·ÛÈÛı› Û ¤Í˘ÓÔ˘˜ ‰È·Ïfi- ÁÔ˘˜ Ù˘ Ì˯·Ó‹˜ Ì ÙÔÓ ¯Ú‹ÛÙË. ÿÚË ÛÙËÓ ÂÚÁ·Û›· Ô˘ ¤ÁÈÓ ÛÙÔ Ï·›ÛÈÔ ÙÔ˘ “¢IA§O°OY” ÚÔÙ›ÓÂÙ·È Ì›· ∆Ô ¤ÚÁÔ “¢IA§O°O™” ›¯Â ÛÙfi¯Ô Ó· ˘Ô‚ÔËı‹ÛÂÈ ÙËÓ Ï‡ÛË ÁÈ· ÙÔ˘˜ ‰È·ÏfiÁÔ˘˜ ÛÙȘ ÂÚÈÙÒÛÂȘ ·ÌÊÈÛËÌ›·˜ ·Ó¿Ù˘ÍË Ù˘ ÁψÛÛÈ΋˜ Ù¯ÓÔÏÔÁ›·˜. ∏ ÊÈÏÔÛÔÊ›· ηٿ ÙËÓ ÌÂÙ¿ÊÚ·ÛË Ì¤Û· ·fi ÙȘ Ù¯ÓÈΤ˜ Ù˘ Á¤ÓÂ- ÙÔ˘ ¤ÚÁÔ˘ ‹Ù·Ó Ë ÂÍ‹˜: Û˘ ÎÂÈ̤ÓÔ˘ (text generation), ÒÛÙ ӷ ˘¿ÚÍÔ˘Ó ‰È¿- ñ ™˘ÁΤÓÙÚˆÛ fiÏÔ˘˜ ÙÔ˘˜ ÂÙ·›ÚÔ˘˜ Ô˘ ›¯·Ó ÁÓÒ- ÏÔÁÔÈ Ô˘ ı· ÂÌÓ¤ÔÓÙ·È ·fi Ù· ÌÂÙ·ÊÚ·˙fiÌÂÓ· Λ- ÛÂȘ, ÂÌÂÈÚ›· Î·È ÂӉȷʤÚÔÓ Ó· ÂÚÁ·ÛıÔ‡Ó ÛÙÔ ÌÂÓ· Î·È ı· Â›Ó·È ¤ÙÛÈ ÂӉȷʤÚÔÓÙ˜ ÁÈ· ÙÔÓ ÌÂÙ·ÊÚ·- Û˘ÁÎÂÎÚÈ̤ÓÔ ÁÓˆÛÙÈÎfi ·ÓÙÈΛÌÂÓÔ ÛÙ‹-¯Ú‹ÛÙË Ô˘ ı· ¯ÚËÛÈÌÔÔÈ› ÙÔ Û‡ÛÙËÌ·. ñ ∂ÈÎÂÓÙÚÒıËΠÛÙËÓ ‰ËÌÈÔ˘ÚÁ›· ·ÚÈıÌÔ‡ ÚÔ˚fiÓÙˆÓ ™ÙËÓ Ì˯·ÓÈ΋ ÌÂÙ¿ÊÚ·ÛË ¤Ó· ‰‡ÛÎÔÏÔ Úfi‚ÏËÌ· ›- ñ ¶ÚÔÂÙԛ̷Û ‚·ÛÈΤ˜ ÌÂϤÙ˜ ÁÈ· ÌÂÏÏÔÓÙÈ΋ Ó·È Û˘Ó‰Â‰Â̤ÓÔ Ì ÙËÓ ˘ÔÏÔÁÈÛÙÈ΋ ÔÏ˘ÏÔÎfiÙËÙ· ¤Ú¢ӷ Î·È ·Ó¿Ù˘ÍË Î·ıÒ˜ Î·È Û˘ÛÙ‹Ì·Ù· Ô˘ Î·È ÙÔÓ ¯ÚfiÓÔ Ô˘ ··ÈÙÂ›Ù·È ÁÈ· ÙËÓ ÌÂÙ¿ÊÚ·ÛË Ì¤- ÂȉÂÈÎÓ‡Ô˘Ó ‰È¿ÊÔÚ˜ ‰˘Ó·ÙfiÙËÙ˜ Ô˘ ÌÔÚÔ‡Ó Û· ·fi ÙËÓ ‰È·‰Èηۛ· Û˘ÓÙ·ÎÙÈ΋˜ ·Ó¿Ï˘Û˘. ™Â Ó· ¯ÚËÛÈÌÔÔÈËıÔ‡Ó Û ÌÂÏÏÔÓÙÈο ÚÔ˚fiÓÙ·. ÔÏϤ˜ ÂÚÈÙÒÛÂȘ Ô ¯ÚfiÓÔ˜ ·˘Ùfi˜ Â›Ó·È ··ÁÔÚ¢- ñ ∂˘·ÈÛıËÙÔÔ›ËÛ Ì ÙÔ ¤ÚÁÔ Ô˘ ¤¯ÂÈ Ó· Âȉ›ÍÂÈ, ÙÈÎfi˜ Î·È ÁÈ· ÙÔÓ ÏfiÁÔ ·˘ÙfiÓ ÂÍÂÙ¿ÛıËÎ·Ó Ï‡ÛÂȘ ·- ÙfiÛÔ ÙÔ˘˜ ÂÙ·›ÚÔ˘˜ Ù˘ ÎÔÈÓÔÚ·Í›·˜ fiÛÔ Î·È Â˘- Ú¿ÏÏËÏ˘ ÂÂÍÂÚÁ·Û›·˜ Ô˘ ÂÈÙ·¯‡ÓÔ˘Ó ÙËÓ fiÏË Ú‡ÙÂÚ· Û¯ÂÙÈο Ì ÙËÓ ÛËÌ·Û›· Ù˘ ÁψÛÛÈ΋˜ ÙÂ- ‰È·‰Èηۛ·. ∞Ó·Ù‡¯ıËΠ̛· ‰˘Ó·ÌÈ΋ Ï·ÙÊfiÚÌ· ¯ÓÔÏÔÁ›·˜ ÁÈ· ÙÔÓ EÏÏËÓÈÎfi ¯ÒÚÔ. ·Ú¿ÏÏËÏ˘ ÂÂÍÂÚÁ·Û›·˜ ȉȷ›ÙÂÚ· ¯Ú‹ÛÈÌË Û ÌÂÏ- ñ ¢ËÌÈÔ‡ÚÁËÛ ‚·ÛÈ΋ ˘Ô‰ÔÌ‹ Û ÊÔÚ›˜ Ô˘ ÙËÓ ÏÔÓÙÈΤ˜ ÚÔÛ¿ıÂȘ ηٷÛ΢‹˜ ‚ÈÔÌ˯·ÓÈÎÒÓ Â›¯·Ó ¿ÌÂÛ· ·Ó¿ÁÎË (πÛÙÔÚÈÎfi §ÂÍÈÎfi Ù˘ ∞η‰Ë- ÚÔÙ‡ˆÓ Ì˯·ÓÈ΋˜ ÌÂÙ¿ÊÚ·Û˘. Ì›·˜ ∞ıËÓÒÓ, ∂ª¶, π∂§). ∆¤ÏÔ˜ Û ۯ¤ÛË Ì ٷ ‰˘Ó·Ù¿ ÂÚÁ·Ï›· ÁÈ· ÙËÓ ‚Ô‹- ıÂÈ· ÙÔ˘ ÌÂÙ·ÊÚ·ÛÙ‹ ÛÙÔ ¤ÚÁÔ ÙÔ˘ ‰fiıËΠ̛· ÛÔ‚·- ºÈÏÔÛÔÊ›· ÙÔ˘ “¢IA§O°OY” Ú‹ ÒıËÛË ÛÙËÓ ·Ó¿Ù˘ÍË ÏÔÁÈÛÌÈÎÔ‡ ÁÈ· ÙËÓ Ï·Ù- √ “¢IA§O°O™” ‹Ù·Ó ¤Ó· ¤ÚÁÔ «ÔÌڤϷ» Ì 13 ÂÙ·›- ÊfiÚÌ· Ù˘ ÌÂÙ·ÊÚ·ÛÙÈ΋˜ ÌÓ‹Ì˘ Ô˘ ‰ËÌÈÔ‡ÚÁËÛ ÚÔ˘˜ Î·È ÌÂÁ¿ÏÔ Â‡ÚÔ˜ ‰Ú·ÛÙËÚÈÔÙ‹ÙˆÓ ÁÈ·Ù› ›¯Â Û·Ó ÙÔ π∂§. √È ÌÂÙ·ÊÚ·ÛÙÈΤ˜ ÌӋ̘ ·ÔÙÂÏÔ‡Ó ÂÍ·ÈÚÂÙÈ- ÛÙfi¯Ô Ó· ηχ„ÂÈ fiÏÔ˘˜ ÙÔ˘˜ ˘ÔÙÔÌ›˜ Ù˘ ÁψÛÛÈ- ΋ χÛË ÁÈ· Ó· ÔÚÁ·ÓÒÛÂÈ Ô ÌÂÙ·ÊÚ·ÛÙ‹˜ ÙÔ ¤ÚÁÔ Î‹˜ Ù¯ÓÔÏÔÁ›·˜ fiÔ˘ ¤Ú ӷ Á›ÓÂÈ ¿ÌÂÛË ÚfiÔ‰Ô˜ ÙÔ˘ Î·È Ó· ¯ÚËÛÈÌÔÔÈ‹ÛÂÈ ÂˆÊÂÏÒ˜ ·Ï·È¤˜ ÔÈÔÙÈ- ÂÓÒ ˘‹Ú¯Â ‹‰Ë οÔÈ· Ù¯ÓÔÁÓˆÛ›·. ¶ÈÔ Û˘ÁÎÂÎÚÈ- Τ˜ ÌÂÙ·ÊÚ¿ÛÂȘ ÙÔ˘ ηٿ ÙËÓ ‰È·‰Èηۛ· ÌÂÙ¿ÊÚ·- ̤ӷ ÙÔ Ê˘ÛÈÎfi ·ÓÙÈΛÌÂÓÔ ÙÔ˘ ¤ÚÁÔ˘ ÌÔÚ› Ó· ÂÚÈ- Û˘. ∂ȉÈÎfiÙÂÚ· ηٿ ÙËÓ ÌÂÙ¿ÊÚ·ÛË ÎÂÈÌ¤ÓˆÓ ‰ÈÔÈ- ÁÚ·Ê› Ì ‚¿ÛË ÙȘ ‰Ú¿ÛÂȘ Ô˘ ·Ó·Ù‡¯ıËÎ·Ó ÛÙÔ˘˜ ÎËÙÈÎÔ‡ Ù‡Ô˘ Ë ÈÛÙfiÙËÙ· Â›Ó·È Ì¤ÚÔ˜ Ù˘ ÔÈfiÙËÙ·˜ ‰È¿ÊÔÚÔ˘˜ ÙÔÌ›˜ ¤Ú¢ӷ˜ Î·È ·Ó¿Ù˘Í˘. ÙÔ˘ ÎÂÈ̤ÓÔ˘ ÛÙËÓ ÁÏÒÛÛ· ÛÙfi¯Ô. √È ÌÂÙ·ÊÚ·ÛÙÈΤ˜ ñ °Ú·ÌÌ‹ ¢Ú¿Û˘ 1: «ªË¯·ÓÈ΋ ªÂÙ¿ÊÚ·ÛË» ÌӋ̘ ÂÍ·ÛÊ·Ï›˙Ô˘Ó ÙËÓ ÈÛÙfiÙËÙ· Î·È ÙËÓ Û˘Ó¤ÂÈ· ñ °Ú·ÌÌ‹ ¢Ú¿Û˘ 2: «∏ÏÂÎÙÚÔÓÈ΋ §ÂÍÈÎÔÁÚ·Ê›·» ηٿ ÙËÓ ÌÂÙ¿ÊÚ·ÛË Ù¤ÙÔÈÔ˘ Ù‡Ô˘ ÎÂÈÌ¤ÓˆÓ Ô˘ ›- ñ °Ú·ÌÌ‹ ¢Ú¿Û˘ 3: «¢ÈÔÚıˆÙ‹˜ ™˘ÓÙ·ÎÙÈÎÒÓ §·ıÒÓ Ó·È ÁÂÓÈο ÁÂÌ¿Ù· ·fi ÛÙÂÚÂfiÙ˘·. To π∂§ Ô‰‹ÁËÛ ÁÈ· ÙËÓ ¡¤· ∂ÏÏËÓÈ΋» Û ÙÂÏÈÎfi ÚÔ˚fiÓ, ÙËÓ Ï·ÙÊfiÚÌ· Ô˘ ›¯Â ·Ó·Ù‡ÍÂÈ ñ °Ú·ÌÌ‹ ¢Ú¿Û˘ 4: «¢È‰·Ûηϛ· Ù˘ ∂ÏÏËÓÈ΋˜ ˆ˜ Û ÚÔÁÂÓ¤ÛÙÂÚ· ¤ÚÁ· ˘fi ÂÚÁ·ÛÙËÚȷ΋ ÌÔÚÊ‹. •¤Ó˘ °ÏÒÛÛ·˜» ñ °Ú·ÌÌ‹ ¢Ú¿Û˘ 5: «™‡ÓıÂÛË ºˆÓ‹˜ ·fi ∫›ÌÂÓÔ» ∏ÏÂÎÙÚÔÓÈ΋ ÏÂÍÈÎÔÁÚ·Ê›· (À‡ı˘ÓÔ˜ π∂§) ñ °Ú·ÌÌ‹ ¢Ú¿Û˘ 6: «¢È·ʤ˜ ºˆÓ‹˜» ™ÙÔÓ ÙÔ̤· ·˘ÙfiÓ fiˆ˜ Î·È ÛÙÔÓ ÚÔËÁÔ‡ÌÂÓÔ ˘‹Ú- ñ °Ú·ÌÌ‹ ¢Ú¿Û˘ 7: «™‡ÓıÂÛË ∫ÂÈ̤ÓÔ˘» Í·Ó ‰Ú¿ÛÂȘ Ì ÂÚ¢ÓËÙÈ΋ ·Í›· fiˆ˜ Î·È ‰Ú¿ÛÂȘ ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 4 / page 4 M¿ÚÙÈÔ˜ 2000 / March 2000
Ô˘ ‹Ù·Ó Û¯ÂÙÈΤ˜ Ì ·Ó¿Ù˘ÍË ÏÔÁÈÛÌÈÎÔ‡, ηıÒ˜ ™˘ÓıÂÙÈ΋ ʈӋ Î·È Û˘ÛÙ‹Ì·Ù· ÁÈ· Ù˘- Î·È ‰Ú¿ÛÂȘ ˘Ô‰ÔÌ‹˜. ÊÏÔ‡˜ (À‡ı˘ÓÔ˜ π∂§) ™Â fiÙÈ ·ÊÔÚ¿ ÙËÓ ˘Ô‰ÔÌ‹ ¤ÁÈÓ Ì ÂÈÙ˘¯›· Ë ÂÈÛ·- ™Ùfi¯Ô˜ ·˘ÙÔ‡ ÙÔ˘ ·Î¤ÙÔ˘ ÂÚÁ·Û›·˜ ‹Ù·Ó Ë ‰ËÌÈÔ˘Ú- ÁˆÁ‹ Ù˘ ÏËÚÔÊÔÚÈ΋˜ ÁÈ· ÙËÓ Î·Ï‡ÙÂÚË ÔÚÁ¿ÓˆÛË Á›· ÂÓfi˜ EÏÏËÓÈÎÔ‡ Û˘Óı¤ÙË Û ·Ú·ı˘ÚÈÎfi ÂÚÈ‚¿Ï- Ù˘ ÂÚÁ·Û›·˜ ÛÙÔ ÈÛÙÔÚÈÎfi ÏÂÍÈÎfi Ù˘ ∞η‰ËÌ›·˜ ∞ıË- ÏÔÓ Î·ıÒ˜ Î·È Ë ¯Ú‹ÛË ÙÔ˘ ÁÈ· ÙËÓ ‰È¢ÎfiÏ˘ÓÛË Ù˘ ÓÒÓ. ∏ ÏÂÍÈÎÔÁÚ·ÊÈ΋ ÂÚÁ·Û›· ÛÙÔÓ ÊÔÚ¤· ·˘ÙfiÓ ·Ó¿ÁÓˆÛ˘ ·fi Ù˘ÊÏÔ‡˜. Œ¯ÂÈ ˘ÏÔÔÈËı› ¤Ó· Û‡- ‹Ù·Ó ÚÔ ÙÔ˘ “¢IA§O°OY” ¯ÂÈÚˆÓ·ÎÙÈ΋. «ªË¯·ÓÔ- ÛÙËÌ· Ô˘ ÂÈÙÚ¤ÂÈ ÙËÓ «ÂÎÊÒÓËÛË» ÎÂÈ̤ÓÔ˘ Ì Ô- ÁÚ·Ê‹ıËλ ÙÔ ÚÒÙÔ Â›Â‰Ô Î·È ÙÔ ‰Â‡ÙÂÚÔ Â›Â- χ ¿ÌÂÛÔ ÙÚfiÔ. ŒÙÛÈ Ô ¯Ú‹ÛÙ˘ ÙÔ˘ ∏/À ı· ¤¯ÂÈ ÙËÓ ‰Ô ηٿÚÙÈÛ˘ ÙÔ˘ ÈÛÙÔÚÈÎÔ‡ ÏÂÍÈÎÔ‡ (ÚˆÙÔÁÂÓ¤˜ ‰˘Ó·ÙfiÙËÙ· Ó· ·ÎÔ‡ÂÈ Ù· ΛÌÂÓ· ·ÓÙ› Ó· Ù· ‰È·‚¿˙ÂÈ Î·È ‰Â˘ÙÂÚÔÁÂÓ¤˜) Ô˘ ‹Ù·Ó Î·È Ù· ϤÔÓ Â›ÁÔÓÙ·, Ú¿ÁÌ· Ô˘ ı· ‚ÂÏÙÈÒÛÂÈ Î·Ù¿ Ôχ ÙËÓ ÔÈfiÙËÙ· ˙ˆ- ÂÓÒ ¯¿ÚË Û ÂȉÈÎfi ¤ÚÁÔ ÙÔ˘ ÀÔ˘ÚÁ›Ԣ ¶·È‰Â›·˜ ı· ‹˜ ÙÔ˘. √È ‰ÔÎÈ̤˜ ÏÂÈÙÔ˘ÚÁ›·˜ ÙÔ˘ Û˘Óı¤ÙË Â›ıÔ˘Ó Ì˯·ÓÔÁÚ·ÊËı› Î·È ÙÔ ÙÚ›ÙÔ Â›Â‰Ô. ™ÙËÓ ˘ÔÏÔÁÈ- fiÙÈ Ë Û˘ÓıÂÙÈ΋ ʈӋ ¤¯ÂÈ ·Ô‰ÂÎÙ‹ ÔÈfiÙËÙ·. ∆fiÛÔ ÛÙÈ΋ ÏÂÍÈÎÔÁÚ·Ê›· ¤¯ÂÈ ÌÂÁ¿ÏË ÛËÌ·Û›· Ó· ˘¿Ú¯ÂÈ ¯Ú‹ÛÙ˜ Ì ÌÂȈ̤ÓË fiÚ·ÛË fiÛÔ Î·È ¯Ú‹ÛÙ˜ ¯ˆÚ›˜ Û˘ÁÎÂÎÚÈ̤ÓÔ ÏÂÍÈÎÔÁÚ·ÊÈÎfi ÌÔÓÙ¤ÏÔ ÁÈ· ÙËÓ ÂÈÙ˘- ÚÔ‚Ï‹Ì·Ù· fiÚ·Û˘ ¤Î·Ó·Ó ıÂÙÈ΋ ·ÍÈÔÏfiÁËÛË ÁÈ· ¯Ë̤ÓË Îˆ‰ÈÎÔÔ›ËÛË Ù˘ ÏËÚÔÊÔÚ›·˜. ™ÙÔ Ï·›ÛÈÔ ÙËÓ ÔÈfiÙËÙ· Ù˘ Û˘ÓıÂÙÈ΋˜ ʈӋ˜. ∂›Û˘ ÙÔ ·Î¤- ÙÔ˘ “¢IA§O°OY” ÚÔÙ¿ıËÎ·Ó ‰‡Ô Ù¤ÙÔÈ· ÌÔÓ٤Ϸ ÙÔ ÂÚÁ·Û›·˜ ·˘Ùfi η٤ÏËÍ Û ÙÂÎÌËÚȈ̤Ó˜ ÚÔÙ¿- Û˘Ì‚·Ù¿ Ì ٷ ‰ÈÂıÓ‹ ÚfiÙ˘· ÁÈ· ÙËÓ Îˆ‰ÈÎÔÔ›ËÛË ÛÂȘ Î·È ÛÂÓ¿ÚÈ· ÁÈ· ÙËÓ ·ÔÙÂÏÂÛÌ·ÙÈ΋ Û˘ÓÂÚÁ·Û›· Ù˘ ÌÔÚÊÔÏÔÁÈ΋˜ ÏËÚÔÊÔÚ›·˜. ∏ ÂÚÁ·Û›· ·˘Ù‹ ›- ÙÔ˘ ÚÔ·Ó·ÊÂÚÔ̤ÓÔ˘ Û˘Óı¤ÙË Î·È ÂÓfi˜ Û˘ÛÙ‹Ì·ÙÔ˜ Ó·È ÌÂÁ¿Ï˘ ÛËÌ·Û›·˜ ÁÈ· ÌÂÏÏÔÓÙÈΤ˜ ÂÓ¤ÚÁÂȘ ÂÓ ÔÙÈ΋˜ ·Ó·ÁÓÒÚÈÛ˘ ¯·Ú·ÎÙ‹ÚˆÓ. fi„ÂÈ ‰ËÌÈÔ˘ÚÁ›·˜ ˘ÔÏÔÁÈÛÙÈÎÒÓ ÏÂÍÈÎÒÓ. ∆¤ÏÔ˜ ÛÙËÓ ‰Ú¿ÛË ·˘Ù‹ ·ÓÂÙ‡¯ıË Î·È ¤Ó· ÚÔ˚fiÓ, ∆ËÏÂÙÚ·Â˙ÈÎfi Û‡ÛÙËÌ· ʈÓËÙÈ΋˜ ¤Ó· ›‰Ô˜ ÏÂÍÈÎÔ‡ ÔÏ˘Ì¤ÛˆÓ ÁÈ· ·È‰È¿ Ô˘ ÈÛÙ‡Â- ÂÈÎÔÈÓˆÓ›·˜ Ù·È fiÙÈ ı· ¤¯ÂÈ ÂÌÔÚÈ΋ ÂÈÙ˘¯›·. ¢ÂÓ ˘‹Ú¯Â ηÏÔ- (À‡ı˘ÓÔ˜ ¶·ÓÂÈÛÙ‹ÌÈÔ ¶·ÙÚÒÓ) ۯ‰ȷṲ̂ÓÔ ÏÂÍÈÎfi ÁÈ· ·È‰È¿ ̤¯ÚÈ Û‹ÌÂÚ· Ô‡Ù Û ∆Ô ·ÓÙÈΛÌÂÓÔ ÙÔ˘ ·Î¤ÙÔ˘ ·˘ÙÔ‡ ÂÚÁ·Û›·˜, ‹Ù·Ó Ë Û˘Ì‚·ÙÈ΋ ÌÔÚÊ‹. ¶ÈÛÙ‡ÂÙ·È fiÙÈ ÙÔ ÏÂÍÈÎfi ·˘Ùfi ı· ‰ËÌÈÔ˘ÚÁ›· ÂÓfi˜ ÚˆÙfiÙ˘Ô˘ ·˘ÙfiÌ·ÙÔ˘ ÚÔÊÔÚÈ- ηχ„ÂÈ ÔÏϤ˜ ·Ó¿ÁΘ. ÎÔ‡ ‰È·ÏÔÁÈÎÔ‡ Û˘ÛÙ‹Ì·ÙÔ˜ (Spoken Dialogue System) ÙÔ ÔÔ›Ô Ó· ¤¯ÂÈ ÙËÓ ‰˘Ó·ÙfiÙËÙ· Ó· Â͢Ë- ¢ÈfiÚıˆÛË Ï·ıÒÓ (À‡ı˘ÓÔ˜ π∂§) ÚÂÙ› ÙÔ˘˜ ÂÏ¿Ù˜ ÙÚ·Â˙ÒÓ fiÏÔ ÙÔ ÂÈÎÔÛÈÙÂÙÚ¿ˆÚÔ ∞ÓÙÈΛÌÂÓÔ ·˘Ù‹˜ Ù˘ ‰Ú¿Û˘ ‹Ù·Ó Ó· ·Ó·Ù˘¯ı› ‰›ÓÔÓÙ·˜ ÏËÚÔÊÔڛ˜ Ô˘ ·ÊÔÚÔ‡Ó Û˘Ó¿ÏÏ·ÁÌ·, ¤Ó·˜ ÔÏÔÎÏËڈ̤ÓÔ˜ ‰ÈÔÚıˆÙ‹˜ Ï·ıÒÓ Ô˘ ÂÎÙfi˜ ÌÂÙÔ¯¤˜ ¯ÚËÌ·ÙÈÛÙËÚ›Ô˘, ˘fiÏÔÈÔ ÏÔÁ·ÚÈ·ÛÌÒÓ ·fi ÔÚıÔÁÚ·ÊÈο Ï¿ıË Ó· ‰ÈÔÚıÒÓÂÈ ·ÚÈıÌfi Ï·ıÒÓ Î.Ï., ‹ ÚÔ‚·›ÓÔÓÙ·˜ Û ÙÚ·Â˙ÈΤ˜ ÎÈÓ‹ÛÂȘ (ÌÂÙ·- Û˘Ìʈӛ·˜ ÌÂٷ͇ ÙˆÓ Ï¤ÍˆÓ. ¢ËÌÈÔ˘ÚÁ‹ıËΠfiÏË Ë ÊÔÚ¿, ÏËڈ̋ ÏÔÁ·ÚÈ·ÛÌÒÓ Î.¿.) ηÙfiÈÓ ·Ú·ÁÁÂ- ˘Ô‰ÔÌ‹ Î·È ·ÓÂÙ‡¯ıË ¤Ó· Ï‹Ú˜ Û‡ÛÙËÌ·. ªÂ ‚¿- Ï›·˜ ÙÔ˘ ÂÏ¿ÙË ¯Ú‹ÛÙË Î·È ÌfiÓÔ ·Ó¿ÌÂÛ· Û ÏÔÁ·- ÛË ÙËÓ Ï·ÙÊfiÚÌ· Ô˘ ¤¯ÂÈ ‰ËÌÈÔ˘ÚÁËı› ·Ó·Ù‡- ÚÈ·ÛÌÔ‡˜ ÙÔ˘ ȉ›Ô˘ ÂÏ¿ÙË. ∆Ô Û‡ÛÙËÌ· Û˘Ó‰¤ÂÙ·È ¯ıËΠ¤Ó· Ó¤Ô ÚÔ˚fiÓ. ÛÙËÓ ÎÂÓÙÚÈ΋ ‚¿ÛË ÙˆÓ ÂÏ·ÙÒÓ Ù˘ ÙÚ¿Â˙·˜ Î·È ÌÔÚ› Ó· ·ÓÙÏ‹ÛÂÈ ‹ Ó· ·ÔıË·ÛÂÈ Û ·˘Ù‹Ó Û˘- ¢È‰·Ûηϛ· Ù˘ EÏÏËÓÈ΋˜ ˆ˜ ͤÓ˘ ÁÎÂÎÚÈ̤ӷ ÛÙÔȯ›·. °È· Ó· ‰È·ÛÊ·ÏÈÛÙ› Ë ·ÛÊ¿ÏÂÈ· ÁÏÒÛÛ·˜ (À‡ı˘ÓÔ˜ π∂§) ÙˆÓ Û˘Ó·ÏÏ·ÁÒÓ ·fi ÙÔ Û‡ÛÙËÌ· Ô ÂÏ¿Ù˘/ ¯Ú‹ÛÙ˘ ™ÙÔÓ ÙÔ̤· ·˘ÙfiÓ ˘‹Ú¯Â ÌÂÁ¿ÏÔ ÎÂÓfi. ◊Ù·Ó ··- ηÏÂ›Ù·È Ó· ÚÔʤÚÂÈ ‰‡Ô Έ‰ÈÎÔ‡˜ ÚÈÓ ¯ÚËÛÈÌÔÔÈ‹- Ú·›ÙËÙÔ Ó· ‰ËÌÈÔ˘ÚÁËı› ¤Ó· ÏÔÁÈÛÌÈÎfi ÔÏ˘Ì¤ÛˆÓ ÛÂÈ ÙÔ Û‡ÛÙËÌ·. ™ÙËÓ Û˘Ó¤¯ÂÈ· ÂÚˆÙ¿Ù·È Ó· ÂÈϤÍÂÈ È‰È·›ÙÂÚ· Û‡Á¯ÚÔÓ˘ Ù¯ÓÔÏÔÁ›·˜ ÁÈ· ÙËÓ ‰È‰·Ûηϛ· ÔÈ· ·fi ÙȘ ÂÈÏÔÁ¤˜ Ô˘ ÚÔÛʤÚÔÓÙ·È ÂÈı˘Ì›. Ù˘ EÏÏËÓÈ΋˜ ˆ˜ ͤÓ˘ ÁÏÒÛÛ·˜. √ ÛÙfi¯Ô˜ Ô˘ ÂÙ¤- ıË ‹Ù·Ó Ó· ÌÔÚ› Ó· ¯ÚËÛÈÌÔÔÈËı› ÙÔ ÏÔÁÈÛÌÈÎfi ÁÈ· ™‡ÓıÂÛË ÎÂÈ̤ÓÔ˘ ÙËÓ ‰È‰·Ûηϛ· ·Ú¯·Ú›ˆÓ Ì ÂÈÎÔÈÓˆÓȷ΋ ̤ıÔ‰Ô. (À‡ı˘ÓÔ˜ Knowledge A.E.) ∂Í’ ·Ú¯‹˜ Ô ÛÙfi¯Ô˜ ‹Ù·Ó Ó· Êı¿ÛÂÈ Ë Ï·ÙÊfiÚÌ· ̤- ∞ÓÙÈΛÌÂÓÔ ÙÔ˘ ¶·Î¤ÙÔ˘ ∂ÚÁ·Û›·˜ 7.1 ‹Ù·Ó Ë ·Ó¿- ¯ÚÈ ÙËÓ ‰ËÌÈÔ˘ÚÁ›· ÙÂÏÈÎÔ‡ ÚÔ˚fiÓÙÔ˜ Ú¿ÁÌ· Ô˘ Ù˘ÍË ÂÓfi˜ Û˘ÛÙ‹Ì·ÙÔ˜ ·˘ÙfiÌ·Ù˘ ηٿ ÙÔ ‰˘Ó·ÙfiÓ, ÂÂÙ‡¯ıË ·ÚÁfiÙÂÚ·. ·Ú·ÁˆÁ‹˜ ÂÌÔÚÈÎÒÓ ÂÈÛÙÔÏÒÓ Ì ÚÔηıÔÚÈṲ̂- ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 5 / page 5 M¿ÚÙÈÔ˜ 2000 / March 2000
ÓÔ ıÂÌ·ÙÔÏfiÁÈÔ Î·È Ì ÙËÓ ·ÏÏËÏ›‰Ú·ÛË, Ê˘ÛÈο, ñ √ÙÈÎfi˜ ‰›ÛÎÔ˜ ÁÈ· ÙËÓ ‰È‰·Ûηϛ· Ù˘ ∂ÏÏËÓÈ΋˜ ÙÔ˘ ¯Ú‹ÛÙË Û ÔÚÈṲ̂ӷ ·Ú¯Èο ÛÙ¿‰È· Ù˘ ‰ËÌÈÔ˘Ú- ˆ˜ ͤÓ˘ ÁÏÒÛÛ·˜ (“ºÈÏÔÁψÛÛ›·”) - À‡ı.: π∂§ ÁÈ΋˜ ‰È‰·Ûηϛ·˜. °È· ÙÔ ÛÎÔfi ·˘Ùfi ·Ú¯Èο Û˘ÁÎÂ- ñ ¢È·‰Èηۛ· “∂ÎÊÒÓËÛ˘ ÎÂÈ̤ӈӔ Î·È Û‡ÛÙËÌ· ÓÙÚÒıËΠ¤Ó·˜ ÛËÌ·ÓÙÈÎfi˜ fiÁÎÔ˜ (corpus) 1.500 Â- ÌÂÙ·ÙÚÔ‹˜ ʈӋ˜ ۠ΛÌÂÓÔ (À‡ı.: π∂§) Ú›Ô˘, ÂÌÔÚÈÎÒÓ ÂÈÛÙÔÏÒÓ-ÚÔÙ‡ˆÓ. ªÂÙ¿ ÙËÓ ñ ∆ËÏÂÙÚ·Â˙ÈÎfi Û‡ÛÙËÌ· ʈÓËÙÈ΋˜ ÂÈÎÔÈÓˆÓ›·˜ Û˘ÛÙËÌ·ÙÈ΋ Î·È ÔχÏ¢ÚË ÂÂÍÂÚÁ·Û›· ·˘ÙÔ‡ ÙÔ˘ (À‡ı.: ¶·ÓÂÈÛÙ‹ÌÈÔ ¶·ÙÚÒÓ) ÛÒÌ·ÙÔ˜ ÙˆÓ ÚÔÙ‡ˆÓ ·Ó·ÊÔÚ¿˜ (ÂÍ·ÁˆÁ‹ ÚˆÙÔ- ñ ∂ÚÁ·ÏÂ›Ô ÊˆÓËÙÈ΋˜ ·fiÛÙ·Û˘ ÁÈ· ÙËÓ ·Ó·˙‹ÙËÛË ÁÂÓÒÓ ¯·Ú·ÎÙËÚÈÛÙÈÎÒÓ, ‰È¿ÚıÚˆÛË, ηÙËÁÔÚÈÔÔ›Ë- ÏËÚÔÊÔÚÈÒÓ ·fi PC Û ‚¿ÛÂȘ ‰Â‰ÔÌ¤ÓˆÓ ÛË Î.Ï.) ÚԤ΢„·Ó ÔÈ ‚·ÛÈΤ˜ ·Ú¿ÌÂÙÚÔÈ Î·È ¿ÍÔ- (À‡ı.: Unixfor) Ó˜ ÙfiÛÔ ÙˆÓ ıÂÌ·ÙÈÎÒÓ ÌÔÓÙ¤ÏˆÓ fiÛÔ Î·È Ù˘ ÁψÛ- ñ ™‡ÛÙËÌ· ¤Í˘Ó˘ ‰ËÌÈÔ˘ÚÁ›·˜ ÂÌÔÚÈÎÒÓ ÂÈÛÙÔÏÒÓ ÛÔÏÔÁÈ΋˜-˘ÊÔÏÔÁÈ΋˜ ‰ÔÌ‹˜ Ô˘ ı· «Ô‰ËÁ‹ÛÔ˘Ó» (À‡ı.: ∫nowlwdge A.E.) ÛÙËÓ Û˘Ó¤¯ÂÈ· ÙÔÓ Û˘Óı¤ÙË ÛÙËÓ ‰ËÌÈÔ˘ÚÁ›· ÙˆÓ ÂÈ- ÛÙÔÏÒÓ. ªÂÙ¿ ÙËÓ ·ÌʛϢÚË Î·È ·ÌÊ›‰ÚÔÌË ÂÍÂÏÈ- ∞ÍÈÔÔ›ËÛË ÙˆÓ ·ÔÙÂÏÂÛÌ¿ÙˆÓ ÙÔ˘ ÎÙÈ΋ ‰È·‰Èηۛ· Û fiÏÔ ÙÔ ‰È¿ÛÙËÌ· ÙÔ˘ ŒÚÁÔ˘, ÙÔ ¢È·ÏfiÁÔ˘ Û‡ÛÙËÌ· Û‹ÌÂÚ· Â›Ó·È Û ı¤ÛË Ó· «ÚÔÙ›ÓÂÈ» ÛÙÔÓ ∆Ô π∂§ ·ÍÈÔÔÈ› Ù· 4 ÚÔ˚fiÓÙ· Ô˘ ‰ËÌÈÔ˘ÚÁ‹ıËÎ·Ó ¯Ú‹ÛÙË Ï‹ÚÂȘ ÂÌÔÚÈΤ˜ ÂÈÛÙÔϤ˜ Û‡Ìʈӷ Ì Ì ¢ı‡ÓË ÙÔ˘. ¶ÚÔ¯ÒÚËÛ ÂÂÓ‰‡ÔÓÙ·˜ ÛÙËÓ ÙÂÏÂÈ- ÙȘ ‚·ÛÈΤ˜ ηÙ¢ı‡ÓÛÂȘ Ô˘ ¤¯Ô˘Ó ‰Ôı› ·fi ÙÔÓ ÔÔ›ËÛË ÙˆÓ ÚˆÙÔÙ‡ˆÓ Ô˘ ·Ó·Ù‡¯ıËÎ·Ó ·fi ÙÔÓ ›‰ÈÔ (·ÔÛÙÔϤ·˜, ·Ú·Ï‹Ù˘, ¯ÚÔÓÔÏÔÁ›·, ı¤Ì·, ¢È¿ÏÔÁÔ. ‡ÊÔ˜ Î.Ï.). ∆Ô Û‡ÛÙËÌ· ¤¯ÂÈ ‹‰Ë ‰Ôı› Û ÂÌÔÚÈΤ˜ √ “ÂÎʈÓËÙ‹˜”, Ë “ÊÈÏÔÁψÛÛ›·” ÙÔ “Tr.AID” Î·È Ë ÂÙ·ÈÚ›˜ Î·È ¯ÚËÌ·ÙÔÈÛÙˆÙÈÎÔ‡˜ ÔÚÁ·ÓÈÛÌÔ‡˜ Ù˘ “Û˘Ìʈӛ·” Â›Ó·È ‰È·ı¤ÛÈÌ·. fiÏ˘, ÙfiÛÔ ÁÈ· ÙËÓ Î·ıËÌÂÚÈÓ‹ Î·È ‰ÔÎÈÌ·ÛÙÈ΋ ¯Ú‹- ÛË fiÛÔ Î·È ÁÈ· ‰È·Ú΋ ¤ÏÂÁ¯Ô Î·È ÂÍ·ÁˆÁ‹ ·Ú·ÙË- ∆Ô π∂§ Û˘Ó¯›˙ÂÈ Ó· ·Ó·Ù‡ÛÛÂÈ Î·È Ó· ‚ÂÏÙÈÒÓÂÈ ÙȘ Ú‹ÛˆÓ,Ì Ôχ ÈηÓÔÔÈËÙÈο ·ÔÙÂϤÛÌ·Ù·. Ù¯ÓÔÏÔÁ›Â˜ ·˘Ù¤˜ ‰ËÌÈÔ˘ÚÁÒÓÙ·˜ Ӥ˜ ÂÊ·ÚÌÔÁ¤˜.
ªÂıÔ‰ÔÏÔÁ›· ÙÔ˘ “¢π∞§√°√À” √È ˘fiÏÔÈÔÈ ÔÚÁ·ÓÈÛÌÔ› Ù˘ ÎÔÈÓÔÚ·Í›·˜ ·ÍÈÔÔÈ- ∞fi ÏÂ˘Ú¿˜ ÌÂıÔ‰ÔÏÔÁ›·˜ Ë ÎÔÈÓÔÚ·Í›· ·ÎÔÏÔ‡ıË- Ô‡Ó Î·È ÂÍÂÏ›ÛÛÔ˘Ó ÙȘ Ù¯ÓÔÏÔÁ›Â˜ Ô˘ ·Ó¤Ù˘Í·Ó Û ٷ ÂÍ‹˜ ‚‹Ì·Ù·: ÛÙÔ Ï·›ÛÈÔ ÙÔ˘ “¢IA§O°OY”. ñ ∆ÂÎÌËÚ›ˆÛË Ù˘ ÁÓÒÛ˘ ÛÙÔ Â› ̤ÚÔ˘˜ ·ÓÙÈΛÌÂÓÔ ñ ∂ÈÏÔÁ‹ Ù˘ ÂȉÈ΋˜ ÌÂıÔ‰ÔÏÔÁ›·˜ Ô˘ Ô‰‹ÁËÛ ™˘ÓÂÚÁ·˙fiÌÂÓÔÈ ÊÔÚ›˜ Ù˘ ÎÔÈÓÔÚ·Í›·˜ ÛÙËÓ ·Ó¿Ù˘ÍË Î¿ı › ̤ÚÔ˘˜ Û˘ÛÙ‹Ì·ÙÔ˜, ÌÂÙ¿ ∂ıÓÈÎfi ªÂÙÛfi‚ÈÔ ¶ÔÏ˘Ù¯ÓÂ›Ô (∂ª¶), πfiÓÈÔ ¶·ÓÂÈ- ·fi ‰ÔÎÈÌ‹ ÂÓ·ÏÏ·ÎÙÈÎÒÓ ÌÂıfi‰ˆÓ. ÛÙ‹ÌÈÔ (π.¶.), ¶·ÓÂÈÛÙ‹ÌÈÔ ∞ıËÓÒÓ (¶.∞.), ¶·ÓÂÈ- ñ ¢ËÌÈÔ˘ÚÁ›· Ï‹ÚÔ˘˜ ÏÂÈÙÔ˘ÚÁÈ΋˜ Ï·ÙÊfiÚÌ·˜ Û ÛÙ‹ÌÈÔ ¶·ÙÚÒÓ (¶.¶.), ∞η‰ËÌ›· ∞ıËÓÒÓ - πÛÙÔÚÈÎfi οı ÂÚ›ÙˆÛË, Û˘Ì‚·Ù‹˜ Ì ÙÔ ·Ú·ı˘ÚÈÎfi Â- §ÂÍÈÎfi Ù˘ ¡¤·˜ ∂ÏÏËÓÈ΋˜, Epsilon Software A.E., ∂Î- ÚÈ‚¿ÏÏÔÓ ·È‰Â˘Ù‹ÚÈ· ¢Ô‡Î· ∞.∂., ¶Ô˘ÏÈ¿‰Ë˜ & ™˘ÓÂÚÁ¿Ù˜ ñ ÀÈÔı¤ÙËÛË Î¿ı ÊÔÚ¿ Ù˘ ·ÏÔ‡ÛÙÂÚ˘ ‰˘Ó·Ù‹˜ ∞.∂., Unixfor ∞.µ.∂.∂., ∆Ú¿Â˙· ¶ÏËÚÔÊÔÚÈÒÓ ∞.∂., ÂÚÁÔÓÔÌ›·˜ ƒ¿‰ÈÔ ∫·Ú·ÁÈ¿ÓÓË ∞.∂.∆.∂., Knowledge A.E. ñ ∂ÌÏÔ΋ ÙˆÓ ÌÂÏÏÔÓÙÈÎÒÓ ¯ÚËÛÙÒÓ ÛÙ· ‰È¿ÊÔÚ· ÛÙ¿‰È· Ù˘ ·Ó¿Ù˘Í˘
¶ÚÔ˚fiÓÙ· Ô˘ ·Ó·Ù‡¯ıËÎ·Ó ∆· ÚÔ˚fiÓÙ· Ô˘ ·Ó·Ù‡¯ıËÎ·Ó Â›Ó·È Ù· ÂÍ‹˜: ñ ™˘ÁÎÚÔÙË̤ÓÔ ÂÚÈ‚¿ÏÏÔÓ ÁÈ· ÌÂÙ·ÊÚ·ÛÙÈο ÂÚ- Á·Ï›· (Tr. AID) - À‡ı.: π∂§ ñ ∂Î·È‰Â˘ÙÈÎfi ÔÏ˘-ÏÂÍÈÎfi (À‡ı.: ∂Î·È‰Â˘Ù‹ÚÈ· ¢√À∫∞™) ñ ™˘ÓÙ·ÎÙÈÎfi˜ ‰ÈÔÚıˆÙ‹˜ Ù˘ ¡¤·˜ ∂ÏÏËÓÈ΋˜ (“™˘Ìʈӛ·”) - À‡ı.: π∂§ ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 6 / page 6 M¿ÚÙÈÔ˜ 2000 / March 2000
¶ƒ√°ƒ∞ªª∞ “™∞¶ºø” (∂¶∂∆ ππ) Ù· ›‰È· Ì ٷ ÚÔËÁÔ‡ÌÂÓ· ÛÙ· ∞ÁÁÏÈο Î·È ÂÈϤÔÓ ÂÚȤ¯ÔÓÙ·È: ¶·ÚÔ˘Û›·ÛË ÛÙËÓ ËÌÂÚ›‰· Ù˘ °°∂∆ ñ ¶ÔχÁψÛÛ· §ÂÍÈο ÂȉÈο ηٷÛ΢·Ṳ̂ӷ ÁÈ· “KÔÈÓˆÓ›· - ¶ÔÏÈÙÈÛÌfi˜ - ¶ÔÏÈÙÈÛÌÈ΋ ÙÔ˘Ú›ÛÙ˜. BÈÔÌ˯·Ó›·” 3 ª·ÚÙ›Ô˘ 2000 ñ ª·ı‹Ì·Ù· EÏÏËÓÈ΋˜ ÁÈ· ·Ú¯·Ú›Ô˘˜.
“™∞¶ºø” (∫·ÈÓÔÙƠ̂˜ / ¶ÚˆÙÔÙ˘›Â˜ / “™∞¶ºø” / M¤ÚÔ˜ Ô˘ ·ÊÔÚ¿ Ù· ¡∏™π∞ ¶ÚˆÙÔÙ˘›Â˜ ˆ˜ Úfi˜ ÙÔ ˘ÏÈÎfi) ∆Ô ¤ÚÁÔ ·˘Ùfi Û˘ÓÂÈÛʤÚÂÈ ÛÙËÓ ·Ó¿‰ÂÈÍË Ù˘ ÈÛÙÔÚÈ- ΋˜ Î·È Ù˘ ÔÏÈÙÈÛÙÈ΋˜ Ê˘ÛÈÔÁӈ̛·˜ ÙˆÓ EÏÏËÓÈÎÒÓ À¿Ú¯Ô˘Ó ‰È·ÊfiÚˆÓ ÂȉÒÓ Î·ÈÓÔÙƠ̂˜ Û οı ÓËÛ›: ÓËÛÈÒÓ, ·Ó·‰ÂÈÎÓ‡ÂÈ ÙÔÓ ÓËÛÈÒÙÈÎÔ ¯ÒÚÔ Û·Ó Ì¤ÚÔ˜ ñ £ÂÌ·ÙÔÏÔÁÈΤ˜ ηÈÓÔÙƠ̂˜ ÙÔ˘ EÏÏËÓÈÎÔ‡ Î·È Î·Ù’ ¤ÎÙ·ÛË ÙÔ˘ E˘Úˆ·˚ÎÔ‡ Ô- ñ ∆¯ÓÔÏÔÁÈΤ˜ ηÈÓÔÙƠ̂˜ (·. ∆· ÚÒÙ· DVD-ROM ÏÈÙÈÛÌÔ‡. EÏÏËÓÈ΋˜ ηٷÛ΢‹˜, ‚. ¶ÚˆÙfiÙ˘Ë Û‡Ó‰ÂÛË ÏËÚÔÊÔÚ›·˜ Î·È Á. ¡¤Â˜ ÌÂıfi‰Ô˘˜ ·Ó¿ÎÙËÛ˘ √È ÛÙfi¯ÔÈ ÙÔ˘ ¤ÚÁÔ˘ Â›Ó·È ÔÈ ÂÍ‹˜: ÏËÚÔÊÔÚÈÒÓ). ñ ¢ËÌÈÔ˘ÚÁ›· ÙˆÓ ÚÒÙˆÓ DVD-ROM EÏÏËÓÈ΋˜ ·- Ú·ÁˆÁ‹˜ ·ÊÈÂÚˆÌ¤ÓˆÓ ÛÙ· EÏÏËÓÈο ¡ËÛÈ¿: √È ÚˆÙÔÙ˘›Â˜ ˆ˜ ÚÔ˜ ÙËÓ ‰ÔÌ‹ ›ӷÈ: ª‡ÎÔÓÔ˜-¢‹ÏÔ˜, ÕÓ‰ÚÔ˜, ¡¿ÍÔ˜, ¶¿ÚÔ˜, ™·ÓÙÔÚ›ÓË, ñ ™‡ÓıÂÛË Ô˘ ‹Á·Û ·fi ¤Ú¢ӷ: πÛÙÔÚ›·, Áˆ- §¤Û‚Ô˜ (Ù¯ÓÔÏÔÁÈÎfi˜ ÛÙfi¯Ô˜) ÁÚ·Ê›·/ÂÚÈ‚¿ÏÏÔÓ, ·Ú¯·ÈÔÏÔÁ›·, ÔÏÈÙÈÛÌfi˜. ñ ¶ÚÔÒıËÛË ÛËÌ·ÓÙÈÎÒÓ EÏÏËÓÈÎÒÓ ÓËÛÈÒÓ ‰ÈÂıÓÒ˜ ñ ™˘ÛÙËÌ·ÙÔÔ›ËÛË Ù˘ ¢fiÌËÛË Ù˘ ¶ÏËÚÔÊÔÚ›·˜, Û ŒÏÏËÓ˜ & ͤÓÔ˘˜ ÙÔ˘Ú›ÛÙ˜ Ô˘ ÂӉȷʤÚÔ- ÔÏ˘Â›Â‰Ë ‰ÔÌ‹. ÓÙ·È ÁÈ· ÙËÓ ÁÓÒÛË ÙÔ˘ ÙfiÔ˘ Ô˘ ÂÈÛΤÙÔÓÙ·È. ñ ∂˘Î·ÈÚ›· ÁÈ· Ì›· ˆÚ·›· Û‡ÓıÂÛË. √È ÚˆÙÔÙ˘›Â˜ ˆ˜ ÚÔ˜ ÙÔ ˘ÏÈÎfi ›ӷÈ: ñ ∞ÔÎ¿Ï˘„Ë ÛÙÔ˘˜ ÙÔÈÎÔ‡˜ ÏËı˘ÛÌÔ‡˜ Ù˘ ÔÏÈ- ÙÈÛÙÈ΋˜ ÙÔ˘˜ ÎÏËÚÔÓÔÌÈ¿˜ (ÌÂÚÈΤ˜ ÊÔÚ¤˜ ͯ·- ñ ¶ÏÔ‡ÛÈÔ ˘ÏÈÎfi (video ·fi Á˘Ú›ÛÌ·Ù·, ÂÈÎfiÓ˜ ·fi Ṳ̂Ó˘). ¶ÚÔÒıËÛË Ù˘ ‰È‰·Ûηϛ·˜ Ù˘ ÙÔÈ΋˜ ʈÙÔÁÚ·Ê‹ÛÂȘ, ÚˆÙfiÙ˘Ë ÌÔ˘ÛÈ΋, ÚˆÙfiÙ˘· ÈÛÙÔÚ›·˜ Ì ȉȷ›ÙÂÚ· ·Ú·ÛÙ·ÙÈÎfi ÙÚfiÔ. ¡· η- ΛÌÂÓ·) Ù·ÛÙ› ÂÓÂÚÁfi οÔÈÔ ·ÓÂÓÂÚÁfi ÈÛÙÔÚÈÎfi ˘ÏÈÎfi. ñ À„ËÏ‹˜ ÔÈfiÙËÙ·˜ ˘ÏÈÎfi (·˘ÛÙËÚ‹ ÂÈÏÔÁ‹) ñ ¶ÚÒÙÔ ¤Ó·˘ÛÌ· Û ÂÚ¢ÓËÙ¤˜ Ô˘ ı¤ÏÔ˘Ó Ó· ·Û¯ÔÏËıÔ‡Ó Ì ÙÔ ·ÓÙÈΛÌÂÓÔ. ∞ÍÈÔÔ›ËÛË ∂ÏÏËÓÈο & •ÂÓfiÁψÛÛ· DVD-ROM ∏ ·ÍÈÔÔ›ËÛË ÙˆÓ ·ÔÙÂÏÂÛÌ¿ÙˆÓ ÙÔ˘ ¤ÚÁÔ˘ fiÛÔÓ ·ÊÔÚ¿ ÙËÓ EÏÏËÓÈ΋ ¤Î‰ÔÛË ı· Á›ÓÂÈ: ·)̤ۈ Ù˘ ñ ¶ÂÚȯfiÌÂÓ· EÏÏËÓÈ΋˜ ¤Î‰ÔÛ˘: ‰È‰·Ûηϛ·˜ Ù˘ ÙÔÈ΋˜ ÈÛÙÔÚ›·˜ ÛÙ· Û¯ÔÏ›· ÙˆÓ ÓËÛÈÒÓ, ‚)¯¿ÚË ÛÙËÓ ‚·ı‡ÙÂÚË ÁÓÒÛË ÙˆÓ ÓËÛÈÒÓ ∆· ÛËÌ›· ÂÈÛfi‰Ô˘ ÁÈ· ÙËÓ EÏÏËÓÈ΋ ¤Î‰ÔÛË Â›Ó·È: ·fi ÙÔÓ EÏÏËÓÈÎfi ÏËı˘ÛÌfi Î·È Á) ¯¿ÚË ÛÙËÓ Î·Ï‡- ñ ªÔ˘ÛÂ›Ô ÙÂÚË ÚÔÛ¤ÁÁÈÛË ÙˆÓ ÓËÛÈÒÓ ·fi ÙÔÓ ŒÏÏËÓ· ÙÔ˘- ñ µÈ‚ÏÈÔı‹ÎË Ú›ÛÙ·. ñ ¶ÂÚÈËÁ‹ÛÂȘ ñ £ÂÌ·ÙÈÎÔ› ¿ÍÔÓ˜ (ÈÛÙÔÚ›·, ÁˆÁÚ·Ê›·/ÂÚÈ‚¿ÏÏÔÓ, ∏ ·ÍÈÔÔ›ËÛË ÙˆÓ ·ÔÙÂÏÂÛÌ¿ÙˆÓ Ù˘ ·ÁÁÏÈ΋˜ ¤Î- ıÚËÛΛ·, Û‡Á¯ÚÔÓË ˙ˆ‹, ·Ú¯·ÈÔÏÔÁ›·, ÔÈÎÔÓÔÌ›·, ‰ÔÛ˘ ı· Á›ÓÂÈ ˆ˜ ÂÍ‹˜: ·. £· ˘¿ÚÍÂÈ ¤Ó· ÂÚÁ·ÏÂ›Ô ·Ú·‰fiÛÂȘ / ÛÙÔȯ›· Ï·˚ÎÔ‡ ÔÏÈÙÈÛÌÔ‡, Ì˘ıÔÏÔ- (sophistique) ÛÙ· ¯¤ÚÈ· ÙÔ˘ ∂√∆ ÁÈ· ÙËÓ ÚÔÒıËÛË Á›·) ÙÔ˘ ˘„ËÏÔ‡ ÂȤ‰Ô˘ ÙÔ˘ÚÈÛÌÔ‡ ÛÙ· EÏÏËÓÈο ÓË- ÛÈ¿, ‚. ™Â Û˘ÓÂÚÁ·Û›· Ì ÙËÓ √Ï˘Ìȷ΋ ∞ÂÚÔÔÚ›· ñ ¶ÂÚȯfiÌÂÓ· ÍÂÓfiÁψÛÛ˘ ¤Î‰ÔÛ˘: Î·È ¿ÏϘ ÌÂÙ·ÊÔÚÈΤ˜ ÂÙ·ÈÚ›˜ Î·È Á. £· ·ÔÙÂϤ- ∆· ÛËÌ›· ÂÈÛfi‰Ô˘ Ù˘ ÍÂÓfiÁψÛÛ˘ ¤Î‰ÔÛ˘ Â›Ó·È ÛÂÈ ¯Ú‹ÛÈÌË ·ÚÔ˘Û›· ÙÔ 2004. ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 7 / page 7 M¿ÚÙÈÔ˜ 2000 / March 2000
“™∞¶ºø” / ª¤ÚÔ˜ Ô˘ ·ÊÔÚ¿ Ù· ∂ÏÏË- FROM GRANADA TO ATHENS ÓÈο ¯ÚfiÓÈ· (“∫§∂√¶∞∆ƒ∞”) √ ÛÙfi¯Ô˜ ·˘ÙÔ‡ ÙÔ˘ ̤ÚÔ˘ ÙÔ˘ ¤ÚÁÔ˘ “™·ÊÒ” Â›Ó·È LREC-2000 (31st MAY - 2nd JUNE 2000) Ë ‰ËÌÈÔ˘ÚÁ›· Ì›·˜ ÚˆÙfiÙ˘Ë˜ Û‡ÓıÂÛ˘ Ì ‰È·ıÂ- ZAPPEION MEGARON, Ì·ÙÈÎfi ¯·Ú·ÎÙ‹Ú· ÁÈ· ˘ÔÛÙ‹ÚÈÍË ÙˆÓ Ì·ıËÌ¿ÙˆÓ ATHENS, GREECE ÔÏÈÙÈÛÌÔ‡ ÙfiÛÔ ÛÙ· EÏÏËÓÈο Û¯ÔÏ›· fiÛÔ Î·È ÛÙ· Despina Scutari Û¯ÔÏ›· Ù˘ ÔÌÔÁ¤ÓÂÈ·˜. ILSP - Liaison Department
¶·È‰·ÁˆÁÈΤ˜ ÚˆÙÔÙ˘›Â˜ √È ·È‰·ÁˆÁÈΤ˜ ÚˆÙÔÙ˘›Â˜ ÙÔ˘ ¤ÚÁÔ˘ Â›Ó·È ÔÈ ·Îfi- OVERVIEW ÏÔ˘ı˜: The Second International Conference on Language ñ ¢È·ıÂÌ·ÙÈ΋ ÚÔÛ¤ÁÁÈÛË Resources was organised with great success at the ñ ∫·ÏÏȤÚÁÂÈ· Ù˘ ‰ÈÂÚ¢ÓËÙÈ΋˜ Ì¿ıËÛ˘. Palacio de Exposiciones y Congresos in Granada, Spain, between the 28th and 30th of May 1998. The ¶ÚfiÎÂÈÙ·È ÁÈ· Ì›· ÂÈÛÎfiËÛË ÙˆÓ ∂ÏÏËÓÈÛÙÈÎÒÓ ÃÚfi- conference was hosted by the University of Granada ÓˆÓ ÛÙËÓ ∞›Á˘ÙÔ Ì ٷ Ì¿ÙÈ· ÙˆÓ ÂȉÈÎÒÓ - ÔÌ¿‰· and was oganised with the support of DG-XIII of the ÈÛÙÔÚÈÎÒÓ. European Commission and the Fundacion Banco Central-Hispano. (∆Ô ˘ÏÈÎfi Û˘ÁÎÂÓÙÚÒıËΠ·fi ÂȉÈ΋ ÔÌ¿‰· ÈÛÙÔÚÈ- ÎÒÓ ÙÔ˘ ¶·È‰·ÁˆÁÈÎÔ‡ πÓÛÙÈÙÔ‡ÙÔ˘ ˆ˜ ÚÔ˚fiÓ ¤Ú¢- Following the widespread recognition and influence Ó·˜, ÂÓÒ Û˘ÌÌÂÙ›¯·Ó Ù· ̤ÏË Ù˘ ÔÌ¿‰·˜ ·˘Ù‹˜ Î·È of the first LREC conference, the Second ÛÙÔÓ Û¯Â‰È·ÛÌfi) International Conference on Language Resources and Evaluation will take place at Zappeion Megaron ™À¡∆∂§∂™∆∂™ ™∞¶º√À™: in Athens, Greece between the 31st of May and the I∂§, ¶π, ∫∂∂§ - ∞∫∞¢∏ªπ∞ ∞£∏¡ø¡, ∂•ø∆∂ƒπ∫√π 2nd of June 2000 (pre-conference workshops will be ™À¡∂ƒ°∞∆∂™. organised on the 29th and 30th of May) and will be hosted by the Institute for Language and Speech Processing (ILSP) and the National Technical University of Athens (NTUA).
LREC-2000 has been initiated by ELRA and is organised in cooperation with other Associations and Consortia, including ACL, ALLC, COCOSDA, ORIENTAL COCOSDA, EAFT, EAGLES, EDR, ELSNET, ESCA, EURALEX, FRANCIL, LDC, PAROLE, TELRI, etc., and with major national and international organisations, including the European Commission - DG XIII, ARPA, NSF, the IC/863 HTRDP Project (China), the National Natural Science Foundation of China, the ICSP Permanent Committee (Korea), The Natural Language Technical committee of JEIDA (Japan), and the Japanese Project for International Coordination in Corpora, Assessment and Labelling.
CONFERENCE AIMS In the framework of the Information Society, the ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 8 / page 8 M¿ÚÙÈÔ˜ 2000 / March 2000 pervasive character of Human Language the US ARPA HLT programme, the EU LE Technologies (HLT) and their relevance to practically programme under R&D framework programmes, all the fields of Information Society Technologies (IST) the Francophone Aupelf-Uref programme and has been widely recognised. others, and in the localisation industry (LISA and LRC), it is still subject to substantial unresolved Two issues are currently considered particularly basic research problems. relevant: 1) the availability of language resources and 2) the methods for the evaluation of resources, The aim of this conference is to provide an technologies and products. Substantial mutual overview of the state of the art, to discuss problems benefits can be expected from addressing these and opportunities, and to exchange information issues through international cooperation. regarding ongoing and planned activities, language resources and their applications. It is The term language resources (LR) refers to sets of also intended to discuss evaluation methodologies language data and descriptions in machine readable and demonstrate evaluation tools, and to explore form, used specifically for building and evaluating possibilities and promote initiatives for natural language and speech algorithms or systems, international cooperation in the areas mentioned for software localisation industries and language above. services, for language enabled information and communication services, for electronic commerce, CONFERENCE TOPICS electronic publishing, language studies, by subject- The following list gives some examples of topics which area specialists and end users. will be addressed at the Conference:
Examples of language resources are written and Issues in the design, construction and use of spoken corpora, computational lexica, grammars, Language Resources (LR) (theoretical & best terminology databases, and basic software tools practices) for the acquisition, preparation, collection, management, customisation and use of these and ñ Guidelines, standards, specifications, and models other resources. for LR ñ Organisational issues in the construction, distribution, The relevance of evaluation for Language and use of LR Engineering is increasingly recognised. This ñ Methods, tools, procedures for the acquisition, creation, involves assessment of the state of the art for a annotation, management, access, distribution, and given technology, measuring the progress use of LR achieved within a programme, comparing different ñ Legal aspects and problems in the construction, approaches to a given problem and choosing the access and use of LR best solution, knowing its advantages and ñ Availability and use of generic vs. task / domain drawbacks, assessment of the availability of specific LR technologies for a given application, product ñ Methods for the extraction and acquisition of knowledge benchmarking, and assessment of user (e.g. terms, lexical information, language modelling) satisfaction. from LR ñ Monolingual and multilingual LR HLT and R&D in language technologies have ñ Multimodal and multimedia LR made important advances in the recent past in ñ LR and the needs/opportunities of the emerging various aspects of both written and spoken multimedia cultural industry language processing. Although the evaluation ñ Industrial production and use of LR paradigm has been studied and used in large ñ Integration of various modalities in LR (spoken, visual, national and international programmes, including gestual, textual) ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 9 / page 9 M¿ÚÙÈÔ˜ 2000 / March 2000
ñ Exploitation of LR in different types of applications CONFERENCE PROGRAMME COMMITTEE (language technology, information retrieval, vocal Nicoletta Calzolari, ILC, Pisa, Italy interfaces, electronic commerce, etc.) George Carayannis, ILSP, Athens, Greece ñ Industrial LR requirements and the community’s Khalid Choukri, ELRA, Paris, France response Harald Hoege, Siemens, Munich, Germany ñ Analysis of user needs for LR Bente Maegaard, CST, Copenhagen, Denmark ñ Mechanisms of LR distribution and marketing Joseph Mariani, LIMSI-CNRS, Orsay, France ñ Economics of LR Antonio Zampolli, Pisa University, Pisa, Italy (conference chair) ñ Customisation and use of LR ñ Research issues relevant for LR LOCAL COMMITTEE ñ Issues in Human Language Technologies evaluation George Babiniotis, University of Athens ñ Evaluation, validation, quality assurance of LR Stelios Bakamidis, ILSP, Athens ñ Benchmarking of systems and products; resources George Carayannis, ILSP, Athens (local committee chair) for benchmarking and evaluation Christophoros Charalambakis, University of Athens ñ Evaluation in written language processing (text retrieval, Nikos Chatzigeorgiou, ILSP, Xanthi terminology extraction, message understanding, Ioannis Dologlou, ILSP, Athens text alignment, machine translation, morphosyntactic Maria Gavrilidou, ILSP, Athens tagging, parsing, semantic tagging, word sense Michael Kopidakis, University of Athens disambiguation, text understanding, summarisation, Stella Markantonatou, ILSP, Athens localisation, etc.) George Papakonstantinou, NTUA, Athens ñ Evaluation in spoken language processing (speech Stelios Piperidis, ILSP, Athens recognition and understanding, voice dictation, oral Gregory Stainhauer, ILSP, Athens dialog, speech synthesis, speech coding, speaker Michael Strintzis, University of Thessaloniki and language recognition, etc.) ñ Evaluation of document processing (document ATHENS recognition, on-line and off-line machine and hand- Athens provides an ideal location for the organisation written character recognition etc.) of the LREC-2000 conference. The city, which has ñ Evaluation of (multimedia) document retrieval and been the capital of Greece since 1834, provides an search systems overall view of the basic periods of Greek history: ñ Evaluation of multimodal systems Ancient, Roman, Byzantine and Modern. In addition to ñ Qualitative and perceptive evaluation the Acropolis and the Parthenon, there are many ñ Evaluation of products and applications beautiful sights such as: the Greek Parliament, the ñ Blackbox, glassbox and diagnostic evaluation of ancient “Herodion” theatre, the marble stadium (the systems first modern Olympic Games were held here in 1896), ñ Situated evaluation of applications Lycabettus Hill (& St. George’s chapel), the old ñ Evaluation methodologies, protocols and measures neighbourhood of Plaka and Monastiraki (traditional ñ From evaluation to standardisation of LR flea market), and the National Garden. Athens enjoys ñ Research issues relevant to evaluation warm, sunny weather during the spring and summer months and allows its visitors to take full advantage of General issues all that it has to offer. Many cultural activities are ñ National and international activities and projects organised in Athens during the summer period in the ñ LR and the needs/opportunities of the emerging framework of the Athens Festival. multimedia cultural industry ñ Priorities, perspectives, strategies in the field of LR CONFERENCE VENUE - ZAPPEION MEGARON national and international policies The conference will take place in the Zappeion ñ Needs, possibilities, forms, initiatives of/for international Megaron which was erected in 1874 – 1888 and was cooperation designed by the architect Theophil Hansen. Zappeion ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 10 / page 10 M¿ÚÙÈÔ˜ 2000 / March 2000
Megaron has witnessed important moments in From Stockholm to Chios Island – Greece’s political history: the Treaty of Accession by the 8th ELSNET European Summer which Greece became a full member of the European School on Language and Speech th Community was signed on its premises on the 29 of Communication May 1979. It normally serves as an international exhibition hall according to the original plan of its (TeSTIA-2000, Chios Island, designers. 15th-30th July, 2000) Anastasios Patrikakos Zappeion Megaron is located in the very center of the ILSP - Liaison Department city, next to the National Garden and just off Syntagma Square, the central square of Athens, and the Parliament. It is within walking distance from the major Since 1993 ELSNET (the European Network in archaeological sites of Athens, the old quarters of the Language and Speech) has organised the annual city and the administrative and business center and, European Summer School in Language and Speech last but not least, close to the central shopping and Communication, better known as the ELSNET recreational areas. The Central Post Office, many Summer School (ESS). banks and a broad range of hotels are also within walking distance. The ESS has become one of the most successful annual training courses in Europe. The target SCIENTIFIC PAPERS audience of the Summer School are advanced About 311 papers wil be published covering many undergraduate students, PhD students, postdocs aspects of Language Resourses and Evaluation. A first and academic and industrial researchers and classification of the papers is as follows: developers. Written 168 Speech 95 Each year the ESS is devoted to a new, advanced Terminology 15 topic in the fields of Language and Speech Evaluation 33 Technology. The topics covered so far are: Prososy Almost 137 papers will be presented by their authors (1993, London, UK), Corpus-Based methods (1994, in oral sessions, while 174 in posters. Utrecht, NL), Multilinguality (1995, Edinburgh, UK), Around 37 papers will include demonstrations. Dialogue Systems (1996, Budapest, HU), Lexicon Development for Language and Speech Processing (1997, Leuven, BE), Robustness: real-life applications in Language and Speech (1998, Barcelona, ES), Multimodality in Language and Speech Systems (1999, Stockholm, SE).
In last year’s ESS in Stockholm multimodality was covered in a multitude of aspects. Conditions and theories for multimodal communication between persons as well as multimodal input/output in technical systems were included in the curriculum.
The programme of the Stockholm ESS included lectures on topics like Multimodal Speech Perception: A Paradigm for Speech Science, Multimodality of meaning in speech and gesture, Multimodality in language and speech systems - from ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 11 / page 11 M¿ÚÙÈÔ˜ 2000 / March 2000 theory to design support tool, Intelligent Multimedia summer school starts every morning with a plenary Presentation Systems, Architectures for integrated session addressed to issues related to the integrated multimodal input-output systems and the humanoid processing of text and speech. Following the plenary interface, Face-to-face communication including session, there are two slots of optional courses and different modalities, Multimodal interaction and practical workshops, run in parallel. In the evenings, people with disabilities, Developing intelligent the students are given the opportunity to present and multimedia applications. discuss their own projects.
The topic selected for 2000 is Text and Speech TeSTIA will be accompanied by a rich cultural Triggered Information Access. Years of speech and programme and many entertainment features, billions of characters are stored in various media including a day trip to the small islands of Inousses, a including the Internet. How can we ever find useful lecture on the history and cultural tradition of the information in such vast archives? Automatic Island which is considered as a possible birthplace of procedures that can recognise speech accurately Homer, a presentation from an art specialist of the and linguistic tools that automatically take out paintings of Chios by Delacroix, etc. essential information components may do the job. The 8th European Summer School on Language and IMPORTANT DATES Speech Communication will present the current state Deadline for pre-registration & grant application: of the art. April 15, 2000 Notification of registration and grants: May 1, 2000 The programme has the following outline: The Payment deadline: June 1, 2000 ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 12 / page 12 M¿ÚÙÈÔ˜ 2000 / March 2000
PROGRAMME COMMITTEE ¶ƒ√°ƒ∞ªª∞∆∞ ª∂∆∞¶∆ÀÃπ∞∫ø¡ Gerrit Bloothooft (Utrecht University, NL) ™¶√À¢ø¡ (∂¶∂∞∂∫) Koenraad de Smedt (University of Bergen, NO) Steve Renals (University of Sheffield, UK) Gregory Grefenstette (Xerox Grenoble, FR) “∆∂á√°§ø™™π∞”: George Carayannis (ILSP, GR) Local org. committee chair ¢ÈÂÈÛÙËÌÔÓÈÎfi – ¢È··ÓÂÈÛÙËÌÈ·Îfi LOCAL ORGANISATION COMMITTEE ¶ÚfiÁÚ·ÌÌ· ªÂÙ·Ù˘¯È·ÎÒÓ ™Ô˘‰ÒÓ Stelios Bakamidis, ILSP, Athens ∞η‰. ∂ÙÔ˜ 1998-1999 George Carayannis, ILSP Athens, Chair Vassilios Digalakis, Technical University of Crete, Chania ∂£¡π∫√ ∫∞π ∫∞¶√¢π™∆ƒπ∞∫√ ¶∞- Ioannis Dologlou, ILSP, Athens ¡∂¶π™∆∏ªπ√ ∞£∏¡ø¡ Panagiotis Konstantopoulos, University of Crete, Iraklio ºÈÏÔÛÔÊÈ΋ ™¯ÔÏ‹, Yannis Kontos, Athens University of Business and ∆Ì‹Ì· ºÈÏÔÏÔÁ›·˜, Economics, Athens ∆Ô̤·˜ °ÏÒÛÛÔÏÔÁ›·˜ George Kouroupetroglou, University of Athens, Athens Ioanna Malagardi, General Secretariat for Research and Technology, Athens ∂£¡π∫√ ª∂∆™√µπ√ ¶√§À∆∂á∂π√ Anastasios Patrikakos, ILSP, Athens ∆Ì‹Ì· ∏ÏÂÎÙÚÔÏfiÁˆÓ ªË¯·ÓÈÎÒÓ Î·È ªË- Stelios Piperidis, ILSP, Athens ¯·ÓÈÎÒÓ ÀÔÏÔÁÈÛÙÒÓ, Athanassios Protopapas, ILSP, Athens ∆Ô̤·˜ ¶ÏËÚÔÊÔÚÈ΋˜ Timoleon Sellis, NTUA, Athens Gregory Stainhauer, ILSP, Athens Athanasios Tsakalidis, Computer Tech. Institute, Patras ∏ ·Ó¿Ù˘ÍË Ù˘ °ÏˆÛÛÈ΋˜ ∆¯ÓÔÏÔÁ›·˜ ··ÈÙ› SPONSORS ÙËÓ ‰ËÌÈÔ˘ÚÁ›· ·Ú¿‰ÔÛ˘ Î·È Ì·ÎÚfiÓÔ˘ ÂÓ·- ELSNET, IHP, and ILSP Û¯fiÏËÛ˘ Î·È ÛÙËÓ ∂ÏÏ¿‰·. ™Â ·˘Ù‹Ó ÙËÓ ·Ó¿ÁÎË Supporting organisations: ESCA and EACL ·ÓÙ·ÔÎÚ›ÓÂÙ·È ÙÔ ÚÔÙÂÈÓfiÌÂÓÔ ¤ÚÁÔ “∆EXNO°§ø™™IA” Ô˘ ·Ó·Ï·Ì‚¿ÓÂÈ Ó· ‰ËÌÈÔ˘Ú- MORE INFORMATION Á‹ÛÂÈ Ì›· ÂÎ·È‰Â˘ÙÈ΋ ‰ÔÌ‹ Ë ÔÔ›· ı· ‚ÔËı‹ÛÂÈ TeSTIA 2000 Secretariat ÛÙËÓ ‰ËÌÈÔ˘ÚÁ›· Ù˘ ÎÚ›ÛÈÌ˘ Ì¿˙·˜ ÂÈÛÙËÌfi- Institute for Language and Speech Processing (ILSP) ÓˆÓ. Artemidos 6 & Epidavrou str. 151 25 Athens, Greece ∏ Û˘Ì‚ÔÏ‹ ÛÙËÓ ·Ó·‚¿ıÌÈÛË ÙˆÓ ÂÎ·È‰Â˘ÙÈÎÒÓ Tel.: +30 1 6875300 ÚÔÁÚ·ÌÌ¿ÙˆÓ, ̤ۈ ÙÔ˘ ¤ÚÁÔ˘ Â›Ó·È ‰Â‰Ô̤ÓË Fax: +30 1 6854270 ‰ÈfiÙÈ Ô ÎÏ¿‰Ô˜ Ù˘ °ÏˆÛÛÈ΋˜ T¯ÓÔÏÔÁ›·˜ ·fi URL: http://www.ilsp.gr/testia/testia2000.html ÙËÓ Ê‡ÛË ÙÔ˘ ¤¯ÂÈ ÙȘ Û˘ÓÈÛÙÒÛ˜: e-mail: [email protected] ·) ÙÔ˘ ÌÂÙ·Ù˘¯È·ÎÔ‡ ÂȤ‰Ô˘ Î·È ‚) Ù˘ ‰ÈÂÈÛÙËÌÔÓÈ΋˜ Û˘ÓÂÚÁ·Û›·˜.
∆Ô ÌÂÙ·Ù˘¯È·Îfi ÚfiÁÚ·ÌÌ· “∆EXNO°§ø™™IA” Â›Ó·È Î·ÈÓÔÙÔÌÈÎfi ÁÈ· ÙÔÓ ∂ÏÏ·‰ÈÎfi ¯ÒÚÔ fi¯È ÌfiÓÔ ˆ˜ Úfi˜ ÙÔ ·ÓÙÈΛÌÂÓÔ ÛÔ˘‰ÒÓ, ·ÏÏ¿ Î·È ˆ˜ Úfi˜ ÙËÓ ÂΠʇÛˆ˜ ·Ó·ÁηÈfiÙËÙ· Û˘ÓÂÚÁ·Û›·˜ ‰‡Ô Û˘ÌÏËڈ̷ÙÈÎÒÓ ÎÏ¿‰ˆÓ ÛÔ˘‰ÒÓ - ÙÔÓ ÎÏ¿‰Ô Ù˘ YÔÏÔÁÈÛÙÈ΋˜ °ÏˆÛÛÔÏÔÁ›·˜ Î·È ÙÔÓ ÎÏ¿‰Ô Ù˘ ¶ÏËÚÔÊÔÚÈ΋˜. ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 13 / page 13 M¿ÚÙÈÔ˜ 2000 / March 2000
¶ÚfiÁÚ·ÌÌ· ™Ô˘‰ÒÓ °’ ∂Í¿ÌËÓÔ
1. ∫·Ù·ÓfiËÛË ∫ÂÈ̤ÓÔ˘ ∞’ ∂Í¿ÌËÓÔ ¢È‰.: ¢Ú. πˆ¿ÓÓ· ª·Ï·Á·Ú‰‹
2. ∆¯ÓÈΤ˜ Parsing ºÈÏfiÏÔÁÔÈ ªË¯·ÓÈÎÔ› ¢È‰.: ∫·ı. °È¿ÓÓ˘ ª·˚ÛÙÚÔ˜ 1. ∂ÈÛ·ÁˆÁ‹ ÛÙÔÓ 1. ∂ÈÛ·ÁˆÁ‹ ÛÙËÓ ÚÔÁÚ·ÌÌ·ÙÈÛÌfi π ÁψÛÛÔÏÔÁ›· 3. ∑ËÙ‹Ì·Ù· ™ËÌ·ÛÈÔÏÔÁÈ΋˜ ∫·ÙËÁÔÚÈÔÔ›ËÛ˘ ¢È‰.: ¢Ú.¡›ÎÔ˜ °Ï¿ÚÔ˜ ¢È‰.: ∫·ıËÁËÙ‹˜ °ÈÒÚÁÔ˜ ÙˆÓ §ÂÍÈÎÒÓ ªÔÓ¿‰ˆÓ ª·ÌÈÓÈÒÙ˘, ¢È‰.: ¢Ú. ª·Ú›· ∆˙‚ÂϤÎÔ˘ – ¶Ô‡ÏÔ˘ Î. ª·ÚÎÔÔ‡ÏÔ˘ 4. ∂ÈÛ·ÁˆÁ‹ ÛÙÔÓ §ÔÁÈÎfi ¶ÚÔÁÚ·ÌÌ·ÙÈÛÌfi 2. ∂ÈÛ·ÁˆÁ‹ ÛÙÔÓ 2. ¢ÔÌ‹ Ù˘ ¡ÂÔÏÏËÓÈ΋˜ ¢È‰.: ∫·ı. °ÈÒÚÁÔ˜ ¶··ÎˆÓÛÙ·ÓÙ›ÓÔ˘, ÚÔÁÚ·ÌÌ·ÙÈÛÌfi ππ °ÏÒÛÛ·˜ ∫·ı. ¶·Ó·ÁÈÒÙ˘ ∆۷ӿη˜, ¢È‰.: ¢Ú. ¡›ÎÔ˜ °Ï¿ÚÔ˜ ¢È‰.: °. ª·ÁÔ˘Ï¿˜, Î. ¡·Ù¿Û· ª·ÓÔ˘ÛÔÔ‡ÏÔ˘ ª. ∫Ô˘ÙÛÔ˘Ï¤ÏÔ˘, ∂. ¶·Ó·Ú¤ÙÔ˘ 5. ªË¯·ÓÈ΋ ªÂÙ¿ÊÚ·ÛË 3. ªË¯·ÓÈÛÌÔ› ·Ú·ÁˆÁ‹˜ 3. ¢ÔÌ‹ ∫ÂÈ̤ÓÔ˘ ¢È‰.: ¢Ú. ∂ϤÓË ∂˘ı˘Ì›Ô˘, ʈӋ˜ Î·È ÙÔ ÊˆÓËÙÈÎfi ¢È‰.: Î.∞ÈÎ.ª·Î¿ÎÔ˘, ¢Ú. ™Ù¤ÏÏ· ª·ÚηÓÙˆÓ¿ÙÔ˘ Û‹Ì· ÛÙÔÓ ∏/À Î.ª.∫·ÎÚȉ‹ ¢È‰.: ∫·ı. °ÈÒÚÁÔ˜ ∫·Ú·ÁÈ¿ÓÓ˘ ¢’ ∂Í¿ÌËÓÔ 4. ∂ÈÎÔÈÓˆÓ›· ·ÓıÚÒÔ˘- 4. ∂ÈÛ·ÁˆÁ‹ ÛÙËÓ Û‡ÓÙ·ÍË Ì˯·Ó‹˜ Î·È ÛËÌ·ÛÈÔÏÔÁ›· 1. ™‡ÓıÂÛË Î·È ∞Ó·ÁÓÒÚÈÛË ºˆÓ‹˜ ¢È‰.: ∫·ı. °ÈÒÚÁÔ˜ ∫·Ú·ÁÈ¿ÓÓ˘ ¢È‰.: Î.£ÂÔÊ·ÓÔÔ‡ÏÔ˘, ¢È‰.: ¢Ú. °ÚËÁfiÚ˘ ™Ù·˚Ó¯¿Ô˘ÂÚ Î. ∞. ªfi˙ÂÚ ¢Ú. ™Ù¤ÏÈÔ˜ ª·Î·Ì›‰Ë˜ 5. °Ú·ÌÌ·ÙÈÎÔ› ºÔÚÌ·ÏÈÛÌÔ› 5. ∂ÈÛ·ÁˆÁ‹ ÛÙËÓ 2. ∞Û·Ê‹˜ §ÔÁÈ΋, °ÂÓÂÙÈÎÔ› ∞ÏÁfiÚÈıÌÔÈ Î·È ÁÈ· ÙËÓ ˘ÔÏÔÁÈÛÙÈ΋ ÌÔÚÊÔÏÔÁÈ΋ ·Ó¿Ï˘ÛË ¡Â˘ÚˆÓÈο ¢›ÎÙ˘· ÁψÛÛÔÏÔÁ›· ¢È‰.: Î.ª·ÚÎfiÔ˘ÏÔ˜, ¢È‰.: Î. ™‡ÚÔ˜ ƒ¿Ù˘ ¢È‰.: ¢Ú. ™Ù¤ÏÏ· ª·ÚηÓÙˆÓ¿ÙÔ˘ Î. ¶. ∫ÔÓÙfi˜ 6. ∂ÈÛ·ÁˆÁ‹ ÛÙËÓ 3. ∂Ê·ÚÌÔÁ¤˜ Ù˘ °ÏˆÛÛÈ΋˜ ∆¯ÓÔÏÔÁ›·˜ Î·È Ù˘ ʈÓËÙÈ΋ Î·È ·ÓÙ›ÏË„Ë ∆¯ÓÔÏÔÁ›·˜ °ÓÒÛ˘ ÛÙËÓ ∞˘ÙÔÌ·ÙÔÔÈË̤ÓË ÔÌÈÏ›·˜ ¶·Ú·ÁˆÁ‹ ∆¯ÓÈÎÒÓ ∫ÂÈÌ¤ÓˆÓ ¢È‰.: Î. ∞. ªÔÙ›Ó˘ ¢È‰.: ¢Ú. ∫ÒÛÙ·˜ ™˘ÚfiÔ˘ÏÔ˜
4. ∞Ó¿ÎÙËÛË ¶ÏËÚÔÊÔÚÈÒÓ µ’ ∂Í¿ÌËÓÔ ¢È‰.: ∫·ı. ∫·Ï·Ìԇ΢ 5. ∞Ú¯ÈÙÂÎÙÔÓÈΤ˜ ™˘ÛÙËÌ¿ÙˆÓ ∂ÂÍÂÚÁ·Û›·˜ 1. ∂ÂÍÂÚÁ·Û›· ™ˆÌ¿ÙˆÓ ∫ÂÈÌ¤ÓˆÓ º˘ÛÈ΋˜ °ÏÒÛÛ·˜ ¢È‰.: ¢Ú. ÿÚ˘ ¶··ÁˆÚÁ›Ô˘ ¢È‰.: Î. ™Ù¤ÏÈÔ˜ ¶ÈÂÚ›‰Ë˜ Î. ™ˆÙ‹Ú˘ ªÔ‡ÙÛ˘
2. §ÔÁÈ΋ Î·È °ÏÒÛÛ· ¢È‰.: Î. ∆¿ÛÔ˜ ¶·ÙÚÈοÎÔ˜ – Î. ™Ù¤ÏÈÔ˜ ¶ÈÂÚ›‰Ë˜
3. ∞Ó¿Ï˘ÛË Î·È ªÔÓÙÂÏÔÔ›ËÛË Ù˘ ºˆÓ‹˜ ¢È‰.: ∫·ı. ¶¤ÙÚÔ˜ ª·Ú·ÁÎfi˜
4. ∂ÈÛ·ÁˆÁ‹ ÛÙËÓ ∏ÏÂÎÙÚÔÓÈ΋ §ÂÍÈÎÔÁÚ·Ê›· ¢È‰.: ∫·ı.ÃÚÈÛÙfiÊÔÚÔ˜ ÷ڷϷ̿΢ Î. ª·Ú›· °·‚ÚÈËÏ›‰Ô˘ ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 14 / page 14 M¿ÚÙÈÔ˜ 2000 / March 2000
TEXT TO SPEECH SYNTHESIS number of parameters to be adjusted, but it actually IN GREEK gave very good results in terms of synthetic voice Dr.Stelios Bakamidis intelligibility and naturalness. The effort was partially ILSP - Department of Speech Technology funded by the “DIALOGOS” project, but the development was continued and finally resulted to a TTS product for the greek language called Speech is by far not only the oldest means of “EKFONITIS” which was finally launched to the greek communication between people but also the most market at the end of 1999 as a CD-ROM for Windows widely used. From the ancient times many attempts 95/98 and Windows NT operating systems. were made towards the construction of machines able to speak but the results were rather “EKFONITIS” is the first TTS product for the Greek disappointing for just two reasons: the complexity of language that appeared in the Greek market and it is the problem and the lack of the necessary seen as a sign of the maturity of the Greek TTS technological background. technology.
With today’s evolution of computers the problem is “EKFONITIS” is a tool that combines a powerful well formulated and we can speak about Text-to- speech synthesis engine with a functional and user Speech Synthesis (TTS) technology that falls into the friendly interface. more general framework of so-called Man-Machine Communication. The general requirement from a TTS “EKFONITIS” offers some special features that are system is the ability to read any electronic text in the described in some detail below: language it has been designed for, in an intelligible and natural manner. FEATURES The TTS conversion is difficult in case we want to produce naturally sounding speech and is still an open ñ Text preprocessing research matter. The high quality TTS synthesis eventually appears in the mid-1980s as a result of EKFONITIS has a module for processing the text prior important advances in speech and natural language to its synthesis. The purpose of this module is to find processing techniques. A TTS system involves many units which cannot be directly read and transcribe modules, the most important of them being the text-to- them to their diction equivalents for the system to be phoneme transcription module, the phonetic units able to read them. Such units include numbers, special database, the prosody generation module and the symbols and punctuation marks. Moreover, this speech synthesizer. module transcribes acronyms or foreign words to greek characters representations according to a table The Institute for Language and Speech Processing of correspondences which can be updated by the user. (ILSP) has gained great experience in TTS due to its years long involvement in the design and ñ Prosody implementation of Greek text to speech synthesis systems. Initially, the speech synthesis research team The term prosody refers to the fluctuation of the tone of ILSP worked on the improvement of the linear and rhythm of the synthetic voice so that it may sound prediction (LP) speech synthesizer platform which had as natural as possible. This description thought is a been developed in the framework of the SPIN project. simplification of a complicated linguistic phenomenon. Soon it was realized that this platform had serious For instance, prosody is responsible for the limitations and another platform was developed based differentiation of positive and negative sentences. on the so-called ‘formant’ synthesizer. The formant EKFONITIS handles non-emphatic positive and synthesizer was more complicated due to the large interrogative sentences. ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 15 / page 15 M¿ÚÙÈÔ˜ 2000 / March 2000
ñ Voice Characteristics modification be extracted and read by “EKFONITIS”. This feature can be used by any application that can move text to “EKFONITIS” offers the user the ability to modify the the clipboard. speed of reading as well as the tone of the synthetic voice. ILSP continues to improve the formant synthesis technology in the framework of the “EYFONON” The speed of reading refers to the rhythm with which project, but at the same time examines other newer “EKFONITIS” speaks, so the reading can be adjusted speech synthesis technologies like the so-called time to be slower or faster. domain techniques which promise according to the international literature naturally sounding synthetic The tone determines the accent that can be low or high. speech in less development time. After comparison, the best synthesis technique will be selected and used in a pilot application which will support reading of e-mail via OPERATION ENVIRONMENT telephone.
“EKFONITIS” is a general purpose tool designed to provide text to speech synthesis services to other text based Windows applications. Special provision has also been made for the automatic incorporation of “EKFONITIS” to the Microsoft Word environment upon its installation.
“EKFONITIS” offers the following mechanisms for text insertion and reading.
ñ Communication with Microsoft Word
Any text loaded to the Microsoft Word environment can be read by “EKFONITIS” by just pressing the hear button appearing in the selection menu after the installation of “EKFONITIS”.
ñ Drag & Drop Operation
Any selected .txt and .doc files can be moved via the well known drag and drop operation to the “EKFONITIS” window for automatic synthesis.
ñ Extensions of the Windows Environment
Any selected .text or .doc file can be read by “EKFONITIS” by simply pressing the right mouse button and left clicking the selection Read.
ñ Copy from Clipboard
The content of the clipboard, if it is in text format, can
∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 16 / page 16 M¿ÚÙÈÔ˜ 2000 / March 2000
SPECTRAL ESTIMATION BASED ON p s (n ) = ™ ·j sn - j (2) THE EIGENANALYSIS i=1 ... OF COMPANION-LIKE MATRICES where aj, j =1, ,q are called prediction coecients Stavroula-Evita Fotinea, which are independent of n. Eq. (2) is exact for q=p in Ioannis Dologlou, Nick Hatzigeorgiu and the absence of noise. In cases where noise is present George Carayannis or the damping is not exponential Eq. (2) is an approximation and it is preferred to chose q much larger than p (over determined model). The new method proposed here is called CSE and March 21, 2000 makes use of a Hankel matrix, a companion matrix based transformation and SVD. The CSE method has Abstract been tested and compared to HTLS, the latter being This paper presents a new state-space method for one of the most promising methods for parameter spectral estimation based on a companion matrix estimation [2]. In the sections that follow the proposed technique in order to estimate frequency, damping CSE method as well as HTLS are presented and the factor, amplitude and phase of exponential sinusoids. superior performance of CSE over HTLS is shown The new method, called CSE, is compared against a through Monte-Carlo based experiments. previously proposed method called HTLS which is based on the use of total least squares. The latter lies among the most promising methods in the field of spectroscopy 2. COMPANION MATRIX BASED where accuracy of parameter estimation is of utmost SPECTRAL ESTIMATION (CSE) importance. Experiments performed on a simulated NMR Let S be the LxM signal observation matrix of a signal prove CSE to be more robust in terms of failure deterministic signal s(n), n = 0,...,N - 1 with rank p < M. rate, especially for low signal to noise ratio. s (0) s (1) . . . s (M - 1) s (1) s (2) . . . s (M ) . . . S = . . . (3) 1. INTRODUCTION . . . Various applications in the field of digital signal ( s (L - 1) s (L) . . . s (N - 1) ) processing, including speech processing [3] as well as spectroscopy, i.e. quantification of NMR signals, with L > p, M >p and L + M - 1 = N are employing exponential sinusoidal models in order Let Sx be the lower shift (top row deleted) to represent a signal segment as a sum of equivalent of S and SX be the upper shift (bottom row exponentially damped complex-valued sinusoids [4]. deleted) equivalent of S. The generalised sinusoidal model we use is given by There is an (L -1)-th order companion matrix C,
p j(Ê0+Êi) (-di+j2fi) n s (n ) = ™ (bie ) e . . . i=1 0 1 0 p (1) . . . n C = . . . = ™ gi zi ,n = O, ..., N - 1 . . . i=1 . . . ( ·1 ·2 · L - 1 ) where p is the model order, i.e. the number of sinusoids that comprise the measured signal. The such that, CSX = Sx (4) objective is to estimate the frequencies fi, damping ... The coecients ot the (L -1)-th row of C depict the factors di, amplitudes bi and phases P0 + Pi, i = 1, , exact linear combination that holds for the rows of SX. p. P0 is the zero order phase, whereas Pi represents extra degrees of freedom. In addition, due to the properties of companion matrices, Linear prediction attempts to estimate a data point the eigenvalues of C are the roots of the polynomial L-1 from a linear combination of past or future ones. For - j ™ ·j z that represent the peaks of the spectrum of the case of forward linear prediction we write j=1 ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 17 / page 17 M¿ÚÙÈÔ˜ 2000 / March 2000 signal s. Consequently, the computation of the least squares sense. Phases and amplitudes are spectral peaks of s may very well be done via the estimated using the least squares method. calculation of the eigenvalues of the companion In detail, the algorithm involves the following four matrix C. steps: In case of real life signals the peaks are embedded in noise and the rank of matrix S is full. The equality (4) Step 1. does not hold any longer for a companion C because We compute the SVD of the Hankel matrix of Eq.(3) the signal does not obey linear models. In this case from the N data points s(n) of Eq.(1): matrix C turns into a companion-like matrix CL where T all the elements are slightly perturbed from their SLXM = ULXL™LXMVMXM theoretical values. The computation of the eigenvalues where L≤M and T devotes the hermitian conjugate. of the companion-like matrix C provides an estimate L The best results are obtained when we use for the spectral peaks of the signal s. When the L=M(+1)=N/2. number of complex peaks to estimate p is known a priori, matrices SX and Sx can be enhanced by Step 2. reducing their rank appropriately. We truncate U,™,V to order p and compute: T To do so we employ the SVD of SX , SX = U™V . To T Sp=Up™pVp where Up, ™p, Vp, are the first p columns of obtain best results L is chosen equal to M(+1) = N/2 T ULXL, ™LXM,VMXM. [1],[2], trying i.e. to use an S as square as possible. Only the p largest singular values are retained to Step 3. enhance SX. The resulting matrix SX e has rank p. We compute the solution Q of Ux p Q = UX p , where Similarly Sx e is computed. Then CL is computed as Ux p(UX p) are derived from Up by deleting its top CL≈ Sx e (SX e )- 1 which results from (4), where the (bottom) row. The eigenvalues Ïi of Q give the signal matrices SX and Sx are replaced by their enhanced pole estimates, which in turn give the estimates for the versions SX and Sx . Note that since both matrices SX e e e damping factors and frequencies of Eq. (1). and Sx e have rank p, CL is also of rank p and this guarantees that only p of the eigenvalues of C are L Step 4. non-zero. The last step is to compute the phases and the In this case, signal pole z estimates are derived i amplitudes. This is done by finding a least squares ... by the eigenvalues Ïi of CL, i = 1, , p. This yields the solution to Eq. (1), with zi replaced by the estimates desired estimates of frequencies f and damping i and s(n) given by the signal data points. factors di from the angles and magnitudes respectively of the eigenvalues of CL. Finally, a computation in a total least squares sense of 4. EXPERIMENTAL RESULTS estimates g follows, where z replaced by their i i Both CSE and HTLS have been tested via estimates. This way, complex-valued linear simulation on a typical NMR signal, in order to parameter estimates of g are calculated, from which i evaluate the improvement in accuracy of parameter amplitude b and phases P + P estimates are i 0 i estimation when using the two methods in the determined as the magnitudes and angles of g i modeling problem defined by Eq.(1). This respectively. representative example simulating a typical 31 P NMR signal of perfused rat liver is presented below. The 31P NMR signal comprises a fifth-order model 3. HTLS method function given in Table 1 by which N data points The HTLS method [2], consists of using the Hankel uniformly sampled at 10KHz are exactly modeled. matrix, performing an SVD decomposition and These data points are perturbed by Gaussian noise reducing the size of matrices to order p. Damping whose real and imaginary components have standard factors d and the frequencies f are estimated in a total i i deviation Ûv. Root mean-squared errors of the
∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 18 / page 18 M¿ÚÙÈÔ˜ 2000 / March 2000 estimates of all signal parameters are computed using The above results suggest that the two methods
500 noise realizations (excluding failures) at each perform similarly for high S/N ratio (Ûv E (0,1)). considered Ûu (when not all peaks are resolved within However, for low S/N ratio (Ûv E (1,2)), despite the specified intervals lying symmetrically around the similarity of the root mean-squared errors of all exact frequencies we consider that a failure has parameters estimated, the CSE method performs occurred). The halfwidths of the intervals are better than HTLS due to its lower failure rate (smaller respectively 82, 82, 82, 43 and 82 Hz, the values number of bad runs compared to HTLS). derived from the Cramer-Rao bounds of peaks 4 and 5 at the noise standard deviation where these intervals are touching each other. The estimated model order is 5. CONCLUSION set to 5. The Cramer-Rao lower bounds are derived In this paper a new state-space method, called from the exact parameter values and Ûv. CSE, for spectral estimation was presented. CSE In Fig. 1 the real part of the fast fourier transform of makes use of a companion-like matrix technique and the simulated 31P NMR signal is depicted. SVD, in order to estimate frequencies, damping factors, amplitudes and phases of exponential peak i fi (Hz) di (rad/s) bi „i(a) sinusoids. CSE was tested in spectroscopy which lies 1 -1379 208 6.1 15 among the most demanding applications of digital 2 -685 256 9.9 15 signal processing in terms of accuracy. CSE was 3 -271 197 6.0 15 compared to the HTLS method, that is the state-of-the- 4 353 117 2.8 15 art method in spectroscopy. A representative example 5 478 808 17.0 15 on a typical 31P NMR signal was presented and the superior performance of CSE over HTLS was shown, (a) „i= P0 Z 180 expresses the phase in degrees especially for low signal to noise ratio.
Table 1: Exact parameter values of the simulated 31P NMR signal, modeled by Eq.(1) with Pi = 0 and number of peaks equalling 5. References Comparative results between CSE and HTLS are presented below for different noise standard [1] I.Dologlou and G.Carayannis, “LPC/SVD analysis deviations. The S/N ratio at peak i is defined by of signals with zero modeling error”, Signal Procesing, 2 2 log10(bi /(2Ûv )). 23 (3), pp. 293-298, 1991. In Fig. 2 failure rates of both CSE and HTLS methods are depicted as a function of noise standard [2] S. van Huffel, H. Chen, C. Decanniere and P. van deviation. Clearly the CSE method is more robust. Hecke, “Algorithm for Time-Domain NMR Data Graphical representation of results does not allow for Fitting Based on Total Least Squares”, J. Magn. Res., a straight forward comparison of the improvement Series A 110, pp. 228-237, 1994. achieved because of the rather small differences in parameter estimation. As seen in Fig. 3 the root mean- [3] R. Kumaresan and D.W. Tufts, “Estimating the squared error is plotted against the S/N ratio achieved parameters of exponentially damped sinusoids and for the frequency estimation of peak 1. The vertical pole-zero modeling in noise”, IEEE Trans.Accoust., dotted line denotes the S/N ratio beyond which the Speech, Signal Proc., ASSP 30 (6), pp. 833-840, badruns of HTLS and CSE methods are the same. To December 1982. clearly present the results, in Table 2 root mean- squared errors of frequency, damping factor, [4] S.Y.Kung, K.S.Arun and D.V. Bhaskar Rao, “Statespace amplitude and phase are depicted respectively for and singular-value decomposition-based approximation peak 1 of the 31P NMR signal for both methods. The methods for the harmonic retrieval problem”, J. Amer. same quantities for peak 4 are presented in Table 3. Opt. Soc., 73 (12), pp. 1799-1811, 1983.
∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 19 / page 19 M¿ÚÙÈÔ˜ 2000 / March 2000
Figure 1: Fast Fourier transform (real part) of the simulated Figure 3: Root mean-squared errors of CSE (.- Z .- ) and 31 ...... P NMR signal. HTLS ( x ) estimates of frequency f1 of peak 1 as a function of the peak S/N ratio which corresponds to a standard deviation Ûv E (0,2).
Ûv/Method f1(Hz) d1(rad/s) b1 „1(deg) 0.1/CSE 0.7782 4.5483 0.0724 1.5442 0.1/HTLS 0.7771 4.5538 0.0726 1.5396 0.3/CSE 2.2771 14.4338 0.2262 4.6035 Figure 2: Percentage of times that CSE (.- Z .- ) and HTLS 0.3/HTLS 2.2812 14.2099 0.2225 4.5913 (...x...) fail to resolve all peaks of the simulated signal using 0.5/CSE 3.8266 23.5800 0.3769 7.5074 N=100 as a function of noise standard deviation Ûu using 0.5/HTLS 3.7338 23.4150 0.3756 7.2648 M = N/2 = 50. 0.7/CSE 5.3719 36.7186 0.5882 10.6468 0.7/HTLS 5.3089 35.4068 0.5689 10.4598 0.9/CSE 7.3128 45.5368 0.7538 14.4414
Ûv/Method f1(Hz) d1(rad/s) b1 „1(deg) 0.9/HTLS 7.2314 42.9739 0.7246 14.0820 0.1/CSE 0.3011 1.8590 0.0328 0.3122 1.1/CSE 9.0421 65.8090 1.0857 18.0785 0.1/HTLS 0.3486 2.0322 0.0377 0.3652 1.1/HTLS 8.9304 56.8974 0.9809 17.5237 0.3/CSE 0.9119 5.5134 0.0989 0.9841 1.3/CSE 10.8140 85.5737 1.4065 21.1541 0.3/HTLS 0.9710 5.8240 0.1334 1.0699 1.3/HTLS 10.6465 65.6707 1.2009 19.9943 0.5/CSE 1.4723 9.4735 0.1680 1.5292 1.5/CSE 13.7314 115.5441 1.9815 26.8592 0.5/HTLS 1.4791 9.5101 0.1739 1.5410 1.5/HTLS 13.8219 76.4961 1.5116 25.8650 0.7/CSE 2.0627 13.4974 0.2559 2.3007 1.7/CSE 14.7433 142.9655 2.3874 28.5107 0.7/HTLS 2.0735 13.3832 0.2515 2.3116 1.7/HTLS 14.3773 94.3041 1.9823 25.2497 0.9/CSE 2.7284 16.7999 0.2956 2.8022 1.9/CSE 17.1385 238.9632 3.3696 32.7830 0.9/HTLS 2.8270 18.7588 0.3641 2.7850 1.9/HTLS 16.9766 108.6047 2.4746 28.6915
1.1/CSE 3.3993 20.3944 0.3607 3.6686 Table 3: Root mean-squared errors of frequency, damping 1.1/HTLS 3.5465 21.2293 0.3874 3.7854 factor, amplitude and phase for peak 4 of the 31P NMR 1.3/CSE 3.9085 25.4938 0.4488 4.0346 signal as a function of noise standard deviation Ûv. The 1.3/HTLS 3.9186 25.1657 0.4432 4.0413 values below the double horizontal line correspond to 1.5/CSE 4.5165 27.9694 0.5193 4.8495 smaller number of bad runs for CSE compared to HTLS, 1.5/HTLS 4.4663 27.1645 0.5136 4.7807 whereas the values above to the same number. 1.7/CSE 5.0495 33.9128 0.5593 5.2325 1.7/HTLS 5.0922 32.4640 0.5552 5.3110 1.9/CSE 5.8908 36.4888 0.6509 5.8916 1.9/HTLS 5.9969 35.0291 0.6613 6.0019
Table 2: Root mean-squared errors of frequency, damping factor, amplitude and phase for peak 1 of the 31P NMR signal as a function of noise standard deviation Ûv. The values below the double horizontal line correspond to smaller number of bad runs for CSE compared to HTLS, whereas the values above to the same number. ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 20 / page 20 M¿ÚÙÈÔ˜ 2000 / March 2000
Sappho Multimedia Bilingual Part of Speech. Dictionaries for Tourists Danai Anagnostopoulou, ñ Examples reflect the sense discrimination of the Elina Desipri, lemmas; the examples listed for each lemma Maria Gavrilidou exemplify its distinguishable meanings. Simple and ILSP - Electronic Lexicography Department understandable examples (with expressions of everyday spoken Greek) have been chosen, which are commonly used by Greek native speakers and The Sappho Multimedia Dictionaries have been which give the typical use of the lemma in specific developed by the Electronic Lexicography Department communicational circumstances. These examples of the Institute for Language and Speech Processing have been selected in order to cover the interests of and consist of a set of bilingual bi-directional different people, who belong to different levels of multimedia dictionaries. Greek language is either language learning. source- or target-language in formulation with one of the following languages: English, French, German, ñ Synonyms correspond to the examples (that is, to Spanish, Russian. the senses of each lemma), due to the fact that every distinguishable sense of the lemma may have Macrostructure of the dictionary a different synonym. Sappho is a set of dictionaries for foreigners, and especially for tourists visiting Greece. Therefore, the ñ Related words list the derivatives of the lemma, lemma list is subcategorised in communicational which are also included in the lemma list. circumstances, which foreigners will most probably encounter. Specifically, it contains 4,082 lemmas, of ñ Category classifies the lemma in one of the which 526 are multiword lemmas (phrases, communicational circumstances within which the expressions, complex nominals etc). The lemma can be encountered. The categories are the methodology adopted for the compilation of the following: lemma list consisted of the following steps: ñ holidays ñ comparative study of several bilingual dictionaries ñ resort as well as dialogs for tourists, in order to extract the ñ nutrition commonly found information included therein, ñ institutions / state ñ creation of a catalog of tourist internet sites and ñ consumer goods critical assessment of the information provided to ñ state of emergency the users ñ public relations ñ extraction of the list of most frequent Greek words ñ economy from the ILSP corpus ñ civilization ñ selection of the final lemma list, from the candidate ñ travel lemma lists compiled from the three above sources. ñ general vocabulary. The basic lemma list has been augmented by catalogs of the most important geographic locations, ñ Subcategory constitutes a more detailed structuring monuments, etc. of the communication circumstances.
Microstructure of the dictionary ñ Translational equivalent into the target language is Every lemma carries the following information: given for each sense. The lemma list and the information provided has ñ Phonetic representation, transcribed according to been translated into the five mentioned languages the International Phonetic Alphabet. (English, French, German, Spanish and Russian) which are combined with Greek. ñ Grammatical category, which defines the lemma The selection of the appropriate translational equivalent ∆‡¯Ô˜ 6 / Issue 6 ñ ÛÂÏ›‰· 21 / page 21 M¿ÚÙÈÔ˜ 2000 / March 2000
was not always straightforward, especially in the case of meanings pertinent to the Greek culture, whose lexical realisation may not exist in the target languages. The problem is that these meanings do not precisely correspond to specific words in every foreign language as is the case with names of typical Greek dishes, Greek dances, instruments etc. The solution provided in these cases was the descriptive definition, and the placement of the specific meaning in context through the example.
The Sappho dictionary is accessible to the user through a multimedia interface, which allows for the search of a word via specific fields of the database (namely lemma, category and subcategory), using Greek either as source or as target language. Thus it caters for two different modes of usage:
(a) the user knows the word in his/her own language and wishes to know the respective Greek word and its usage, or
(b) the user comes across a Greek word whose meaning s/he ignores and wants to look it up in the dictionary.
√‰ËÁ›Â˜ ˘Ô‚ÔÏ‹˜ ¿ÚıÚˆÓ ÚÔ˜ ‰ËÌÔÛ›Â˘ÛË Instructions to Authors
∆· ÚÔ˜ ‰ËÌÔÛ›Â˘ÛË ¿ÚıÚ· Ú¤ÂÈ Ó· ˘Ô‚¿ÏÏÔ- Contributions should be submitted in 3 hard copies to: ÓÙ·È Û ÙÚ›· ·ÓÙ›Ù˘· ÛÙË ‰È‡ı˘ÓÛË: ILSP-Institute for Language and Speech Processing π∂§-πÓÛÙÈÙÔ‡ÙÔ ∂ÂÍÂÚÁ·Û›·˜ ÙÔ˘ §fiÁÔ˘ 6 Artemidos & Epidaurou str., ∞ÚÙ¤ÌȉԘ 6 & ∂ȉ·‡ÚÔ˘, 151 25 Paradisos Amarousiou 151 25 ¶·Ú¿‰ÂÈÛÔ˜ ∞Ì·ÚÔ˘Û›Ô˘ and by e-mail at: [email protected] with the indication Î·È Û ËÏÂÎÙÚÔÓÈ΋ ÌÔÚÊ‹ ÛÙË ‰È‡ı˘ÓÛË: Logonavigation in the Subject line. [email protected] Ì ÙËÓ ¤Ó‰ÂÈÍË §ÔÁÔÏÔ‹ÁËÛË ˆ˜ ı¤Ì·. Articles should be submitted as Word documents, in Times New Roman, 12 pt with double spacing. ∆· ΛÌÂÓ· Ú¤ÂÈ Ó· ηٷٛıÂÓÙ·È Û ÌÔÚÊ‹ Word- ·Ú¯Â›Ô˘ ‰›ÛÙËÏ˘ ÛÂÏ›‰·˜, Ì ¯·Ú·ÎÙ‹Ú˜ Times All graphs must be submitted in .tif format or in any New Roman, ÙˆÓ 12 ÛÙÈÁÌÒÓ Û ‰ÈÏfi ‰È¿ÛÙËÌ·. other high resolution 300 dpi form.
ŸÏ· Ù· ÁÚ·Ê‹Ì·Ù· Ú¤ÂÈ Ó· ηٷٛıÂÓÙ·È Û ÌÔÚ- References should follow the cited pattern in Times Ê‹ .tif ·Ú¯Â›Ô˘ ‹ Û ÔÔÈ·‰‹ÔÙ ¿ÏÏË ÌÔÚÊ‹ ˘„Ë- New Roman, 11 pt: Ï‹˜ ·Ó¿Ï˘Û˘ (high resolution 300 dpi). Daille, B., E. Gaussier & J.M. Lange (1994). Towards automatic extraction of monolingual H ‚È‚ÏÈÔÁÚ·Ê›· Ú¤ÂÈ Ó· ·ÎÔÏÔ˘ı› ÙÔ Û¯‹Ì· Ô˘ and bilingual terminology. In Proceedings of ·Ú·Ù›ıÂÙ·È ˆ˜ ˘fi‰ÂÈÁÌ· Û Times New Roman ¯·- COLING 94, 515_521. Jackendoff, R. (1996). Ú·ÎÙ‹Ú˜ ÙˆÓ 11 ÛÙÈÁÌÒÓ: Languages of the Mind. The MIT Press. Daille, B., E. Gaussier & J.M. Lange (1994). References inside the body text should have the Towards automatic extraction of monolingual form i.e. (Jackendoff, 1996) or (Jackendoff, 1996; and bilingual terminology. In Proceedings of Daille et al, 1994). COLING 94, 515_521. Jackendoff, R. (1996). Languages of the Mind. The MIT Press. Footnotes are placed at the bottom of each page in ™ÙÔ Î›ÌÂÓÔ ÔÈ ‚È‚ÏÈÔÁÚ·ÊÈΤ˜ ·Ó·ÊÔÚ¤˜ ÙÔÔıÂ- Arial Greek, 9 pt. ÙÔ‡ÓÙ·È Û ·ÚÂÓı¤ÛÂȘ Ì ÙÔ fiÓÔÌ· ÙÔ˘ Û˘ÁÁÚ·- ʤ· Î·È ÙË ¯ÚÔÓÔÏÔÁ›· ¤Î‰ÔÛ˘ .¯. (Jackendoff, An abstract of 200 words in Times New Roman, 10 pt, 1996) ‹ ·Ó ·Ú·Ù›ıÂÓÙ·È ÂÚÈÛÛfiÙÂÚ˜ ·fi Ì›· precedes the body text. ·Ó·ÊÔÚ¤˜ (Jackendoff, 1996; Daille et al, 1994). Submissions should not exceed 10 pages (A4) √È ˘ÔÛËÌÂÈÒÛÂȘ ÙÔÔıÂÙÔ‡ÓÙ·È ÛÙÔ Ù¤ÏÔ˜ οı ÛÂÏ›‰·˜ Ì ¯·Ú·ÎÙ‹Ú˜ Arial Greek ÙˆÓ 9 ÛÙÈÁÌÒÓ. Every submission should be accompanied by a note with the name, address, e-mail address, telephone ¶ÚÈÓ ·fi ÙÔ ‚·ÛÈÎfi ΛÌÂÓÔ ÚÔËÁÂ›Ù·È ÂÚ›ÏË„Ë and fax number of the main author. 200 ϤÍÂˆÓ Û Times New Roman ¯·Ú·ÎÙ‹Ú˜ ÙˆÓ 10 ÛÙÈÁÌÒÓ. Articles submitted to Logonavigation should not be published elsewhere before. ∏ οı ˘Ô‚ÔÏ‹ ‰ÂÓ Ú¤ÂÈ Ó· ˘ÂÚ‚·›ÓÂÈ ÙȘ 10 ÛÂÏ›‰Â˜ ∞4.
∫¿ı ˘Ô‚ÔÏ‹ Ú¤ÂÈ Ó· Û˘Óԉ‡ÂÙ·È ·fi ÛËÌ›ˆÌ· Ì ÙÔ fiÓÔÌ· Î·È ÙË ‰È‡ı˘ÓÛË, ÙËÓ ËÏÂÎÙÚÔÓÈ΋ ‰È‡- ı˘ÓÛË, ÙËϤʈÓÔ Î·È fax ÙÔ˘ ‚·ÛÈÎÔ‡ Û˘ÁÁڷʤ·.
∆· ¿ÚıÚ· Ô˘ ˘Ô‚¿ÏÏÔÓÙ·È ÚÔ˜ ‰ËÌÔÛ›Â˘ÛË ÛÙË §ÔÁÔÏÔ‹ÁËÛË ‰ÂÓ Ú¤ÂÈ Ó· ¤¯Ô˘Ó ‰ËÌÔÛÈ¢ı› ÚÔËÁÔ˘Ì¤Óˆ˜ Û ¿ÏÏÔ ¤ÓÙ˘Ô.