Dialogiagenttien toiminta ja suomenkielinen toteutus

Dialogiagentit ovat järjestelmiä, joiden avulla ihminen pystyy keskustelemaan tietokoneen kanssa puheen välityksellä. Puheen avulla ihminen voi antaa järjestelmälle käskyjä, kysyä tietoa tai yksinkertaisesti keskustella luonnollisella tavalla lähes asiasta kuin asiasta. Toimiakseen oikein dialogiagentin on ensin purettava käyttäjän puhe tietokoneelle käytettävään muotoon. Sanat on tunnistettava oikein ja niille on määritettävä merkitys. Vasta tämän jälkeen voidaan hakea haluttu tieto tai suorittaa käsky. Jos tehtävä vaatii vastauksen käyttäjälle, se tuotetaan käyttämällä luonnollisen kielen käsittelyn tekniikoita. Tämä vastaus voidaan joko esittää tekstimuodossa tai syntetisoida puheeksi. Tapoja toteuttaa tällainen järjestelmä on monia. Perinteisemmät järjestelmät rakennetaan tyypillisesti yhtä käyttötarkoitusta varten, mutta uusimmat järjestelmät pyrkivät geneerisyyteen. Monet uusimmista menetelmistä nojautuvat koneoppimiseen. Panostamalla uudelleenkäytettävyyteen ja skaalautuvuuteen järjestelmiä ei tarvitse luoda uudelleen, jos käyttötarkoitusta halutaan vaihtaa. Agentteja on luotu runsaasti englannin kielelle, mutta suomen kielen ominaisuudet luovat erityisiä haasteita järjestelmän kehittämiselle. Suomen kielen lukuisat sijamuodot estävät kaikenkattavien sanastojen käyttämisen ja lähes vapaa sanajärjestys tekee viestien tulkitsemisesta hankalaa. Suomen kieliteknologian resurssit ovat viime vuosina kehittyneet tehokkaasti tästä huolimatta. Tutkielmassa tarkastellaan, missä dialogiagentteja käytetään ja millaisista komponenteista järjestelmät koostuvat. Käydään läpi myös, millaisia haasteita järjestelmän kehittämisessä on. Lopuksi tehdään katsaus suomen kielen tilaan kieliteknologian saralla ja millaisia resursseja suomenkielisen agentin kehittämiseen on.

Quantum wire with periodic serial structure

Electron wave motion in a quantum wire with periodic structure is treated by direct solution of the Schrödinger equation as a mode-matching problem. Our method is particularly useful for a wire consisting of several distinct units, where the total transfer matrix for wave propagation is just the product of those for its basic units. It is generally applicable to any linearly connected serial device, and it can be implemented on a small computer. The one-dimensional mesoscopic crystal recently considered by Ulloa, Castaño, and Kirczenow [Phys. Rev. B 41, 12 350 (1990)] is discussed with our method, and is shown to be a strictly one-dimensional problem. Electron motion in the multiple-stub T-shaped potential well considered by Sols et al. [J. Appl. Phys. 66, 3892 (1989)] is also treated. A structure combining features of both of these is investigated

Fine wire resistance thermometer amplifier for atmospheric measurements

An improved amplifier for atmospheric fine wire resistance thermometry is described. The amplifier uses a low excitation current (50 mu A). This is shown to ensure negligible self-heating of the low mass fine wire resistance sensor, compared with measured nocturnal surface air temperature fluctuations. The system provides sufficient amplification for a +/- 50 degrees C span using a +/- 5 V dynamic range analog-to-digital converter, with a noise level of less than 0.01 degrees C. A Kelvin four-wire connection cancels the effect of long lead resistances: a 50 m length of screened cable connecting the Reading design of fine wire thermometer to the amplifier produced no measurable temperature change at 12 bit resolution.

Comparison of white metal and resin socket terminations for wire ropes

Previously the authors have presented both theoretical and experimental work discussing the operating mechanism of a wire rope held in a tapered socket by means of a cast resin cone. The work reported here extends the investigation to address the question of whether the same socket fabricated with white metal operates in the same manner. To date, previous investigations have compared the operational efficiency of resin and white metal in terms of both strength and/or fatigue endurance. Some other work has analysed the operation of resin sockets or specific cast metal terminations. This paper seeks to draw the results from this work together, and, in addition to a theoretical analysis, presents experimental data obtained from a direct comparison of the operation mechanism for the same sockets filled with resin or white metal. Results show that white metal terminations have a very different distribution of stresses along the length of the socket basket from resin terminations, and a smaller but still significant amount of socket draw. For both types of termination the socket draw develops high frictional gripping forces which can transfer the load from the rope to the socket. The different stress distributions mean that the consequences of termination fabrication defects may not be the same for resin and white metal terminations.

Modelling the torsional interaction of wire and polyester fibre ropes used for offshore moorings

Oil rig mooring lines have traditionally consisted of chain and wire rope. As production has moved into deeper water it has proved advantageous to incorporate sections of fibre rope into the mooring lines. However, this has highlighted torsional interaction problems that can occur when ropes of different types are joined together. This paper describes a method by which the torsional properties of ropes can be modelled and can then be used to calculate the rotation and torque for two ropes connected in series. The method uses numerical representations of the torsional characteristics of both the ropes, and equates the torque generated in each rope under load to determine the rotation at the connection point. Data from rope torsional characterization tests have been analysed to derive constants used in the numerical model. Constants are presented for: a six-strand wire rope; a torque-balanced fibre rope; and a fibre rope that has been designed to be torque-matched to stranded wire rope. The calculation method has been verified by comparing predicted rotations with measured test values. Worked examples are given for a six-strand wire rope connected, firstly, to a torque-balanced fibre rope that offers little rotational restraint, and, secondly, to a fibre rope whose torsional properties are matched to that of the wire rope.

Proposal of redundant drive wire mechanism for producing motions with high acceleration and high precision

A new wire mechanism called Redundant Drive Wire Mechanism (RDWM) is proposed. The purpose of this paper is to build up the theory of a RDWM with fast motion and fine motion. First, the basic concepts of the proposed mechanism is presented. Second, the vector closure condition for the proposed mechanism is developed. Next, we present the basic equations, propose the basic structure of RDWM with the Internal DOF module, Double Actuation Modules and Precision Modules together with the properties of the mechanism. Finally, we conduct the simulation to show the validity of the RDWM.

Kinematical analysis of redundant drive wire mechanisms with velocity constraint

In this paper, we propose a new velocity constraint type for Redundant Drive Wire Mechanisms. The purpose of this paper is to demonstrate that the proposed velocity constraint module can fix the orientation of the movable part and to use the kinematical analysis method to obtain the moving direction of the movable part. First, we discuss the necessity of using this velocity constraint type and the possible applications of the proposed mechanism. Second, we derive the basic equations of a wire mechanism with this constraint type. Next, we present a method of motion analysis on active and passive constraint spaces, which is used to find the moving direction of a movable part. Finally, we apply the above analysis method on a wire mechanism with a velocity constraint module and on a wire mechanism with four double actuator modules. By evaluating the results, we prove the validity of the proposed constraint type.