Language, ablity to speak & write and communicate is one of the most fundamental aspects of human behaviour. As the study of human-languages developed the concept of communicating with non-human devices was investigated. This is the origin of natural language processing (NLP). The idea of natural language processing is to design and build a computer system that will analyze , understand and generate natural human-languages.
Natural language communication with computers has long been a major goal of artificial intelligence, both for the information it can give about intelligence in general, and for practical utility. There are many applications of natural language processing developed over the years. They can be mainly divided into two parts as follows. •Text-based applications This envolves applications such as searching for a certain topic or a keyword in a data base, extracting information from a large document, translating one language to another or summarizing text for different purposes. •Dialogue based applications
Some of the typical examples of this are answering systems that can answer questions, services that can be provided over a telephone without an operator, teaching systems, voice controled mechines (that take instructions by speech) and general problem solving systems. Natural Language Processing is a technique where mechine can become more human and there by reducing the distance between human being and the mechine can bereduced. Therefore in sinple sence NLP makes human to communicate withthe mechine easily. There are many applications developed in past few decades in NLP.
Most of these are very useful in everyday life for example a mechine that takes intructions by voice. There are lots of research groups working on this topic to develope more practical are useful systems. Since the invention of the typewriter, the keyboard has been the king of human-computer interface, largely because it has been the only one widely available. The search for an alternative method, such as speech, has continued since the 1950s and computers that can be voice conrolled have featured in a number of science fiction films, such as ‘2001:A Space
Odyssey’. While the conversational skills of the film’s master computer HAL yet to be mastered by today’s computers, speech recognition technology has made significant advances from the limited systems that began emerging in the late 1980s. At last speech recognition technology is a viable tool for both business and home computer users. Today there are many systems developed using speech recognition technology. Some of them are listed below. VoiceType Dictation
IBM’s VoiceType is a speech recognition system that analyses spoken words and instantly turn them into text on a PC screen, at a typical dictation speeds of 70-100 words per minute with accuracy in access of 90%. It allows users to have hands and eyes free and to talk, rather than type. This works by analysing phonemes – the sound that make up spoken words – and then matching them against the phonemes of words held in the systems’s vocabulary. This process is made more difficult if words are run together in continuous or nornal, everyday speech.
When the phonemes run together rather than been separated into identifiable words by a single pause, a large amount of processing power is needed in order to decipher these complicated patterns. For this reason and for viable speech recognition systems that will run on a desktop or laptop pc, users must speak in discrete speech: that is, they must leave approximately one-tenth of a second between words. IBM launched VoiceType Dictation 3. 0 in june 1996 and is available in the market for under 100 pounds. Some of examples of people who users this are translators, doctors and writers as shown in the above pictures.
BBN Coperation BBN engage in research, development and custom application services in speech and languages. Their major projects include speech access to millatary logistics systems, voice-activated searching for information on the world wide web (using conventional browsers and search engines), spoken interfaces to complex systems, technologies for extracting named entries and simple relationships among those entries from plain text, speech recognition over the internet and speaker authentication for security applications. Sri International There are series of DARPA-funded projects ar SRI International with the goal f creating a technology for understanding spontancous spoken natural languages. This tchnology combines speech recognition with nutural language understanding.
Spoken Language Systems Spoken language systems perform speech recognition and semantic analysis, attempting to understand users’ speech and respond appropriately. SRI’s technology provides high accuracy,and real-time performance for large vocabulary tasks. SRI has developed a spoken language system in the air travel planning domain which permits users to ask naturally phrased questions such as “Show me the cheapest flight from Denver to Boston on Saturday after ten in the morning. The system connects to the Official Airline Guide on-line database and provides current information. Some of the other projects done at here are Neural Network/Hidden Markov Hybrid System, Consistency Modeling, Large Vocabulary Wordspotting, Noise/Channel Robustness, Speech Disfluencies, Voice-Interactive Language Instruction and Evaluation, Speech Machine Translation, Speaker Recognition, Data Collection and Annotation, and Voice Banking. Dragon System Naturally Speaking Dragon NaturallySpeaking is the natural way to input text.
Almost anyone business professionals, writers and journalists, nontypists, telecommuters, and employees of small offices can find themselves quickly creating documents and reports with ease and accuracy. Dragon NaturallySpeaking spells correctly every time. Vocalis Products •The world’s first virtual operator -Operetta Operetta™, from Vocalis, is a new and revolutionary call answering system that can handle your incoming calls 24 hours a day, 365 days a year. Based on advanced speech recognition technology, Operetta offers the highest level of customer service consistently.
This system allows people to interact entirely by voice, with a minimum of fuss. Callers simply say the name of the person or department they want and Operetta does the rest. If the caller doesn’t know who to speak to, Operetta can easily route them to an operator; or if someone is away from their desk, the caller can leave a voice mail message. To select any option the caller just tells the computer, exactly as they would a receptionist. Natural language accessible theatre infomation and booking system. NLP for Air Travel information for customers.
NLP in a spoken language interface for battlefield simulation. This is another area where applications of natural language processing can be seen to extract information required from a large database. There are lots of places where this technique is applied to get things we need faster. Some of them are mentioned below. NetOwl extractor NetOwl extractor is an automatic indexing software that identifies names and other important concepts so that users can quickly locate and access the resources they need.
The NetOwl extractor technology is particularly appealing to content providers and other organisations who need to add this capability to their own search services. This software makes it possible to recognize and process the names and other special artifacts that trip up most text processing software. Basically this is a data extracting engine that identifies and interprets key elements of free text, particularly names of people, places and organizations IBM, Thomson corporation and US Government are some people who use this software. Lingsoft’s Tools for Indexing And Retrieval
In English, no word has more than a handful of inflectional forms. For instance, the verb walk has four forms: walk, walks, walking and walked. That is why traditional indexing programs designed for English have completely ignored morphology. On the other hand, most other European languages have more complex morphology. A Finnish word may appear in hundreds or thousands of different forms. At each word, your indexing program should store both the word itself and all possible base forms in the index. The base forms are usually stored in a separate field to enable both exact and morphological matches.
For instance, when the indexing program encounters the word thought, it should store thought in the exact-match field and the list (think, thought) in the base form field. The retrieval program should work as follows •Get a search term (base form), e. g. think or thought. •Find the records where one of the possible base forms matches the search term. For example, think matches the records think, thinks and thought, while thought matches thought and thoughts. This is one of the most important applications of Natural Language Processing.
Translation of a sentence from one language to another, retaining the meaning, is a difficult task. A lot of reasearch has been done on this now in different parts of the world. Some of the procucts available of this type are as following. Language translators from Globalink •Globalink Web Translator If you spend a lot of time browsing the Web, this product is for you! With Globalink Web Translator, you can translate foreign-language Web sites from Spanish, French or German into English–or vice versa. See what everyone’s talking about! •Globalink Power Translator 6. If your business brings you in contact with foreign languages, this product is suited to help you with all your business communications. Globalink Power Translator 6. 0 has four languages in one box–Spanish, French, German and Italian–so you can easily translate text to and from English. Includes a utility for translating within e-mail applications, a special version of Globalink Web Translator® , and a Conversation utility. The product is based on Globalink Barcelona™ technology and works only with Windows 95 or NT systems. •Globalink Subject Dictionaries
The Subject Dictionary CD is for you! This CD contains over 31 subject dictionaries in Spanish, French, and German and are based on Barcelona technology for use with this product. •Globalink Language Assistant Globalink Language Assistant products are designed for personal use and provide instant help when writing, studying, or translating. The program has extensive grammar help, including complete conjugations for verbs and bilingual dictionaries. This product is for the “casual” translator, using the product for personal and learning applications. •Globalink Talk to Me
Talk to Me is the on interactive language learning program that does both: teaches you to listen and to speak. Talk to Me is the self-paced way to learn a foreign language Applications These are some other organizations that have developed NLP applications. CoGenTex Inc CoGen Tex Inc specialises in the developement of softeware systems which represent practical applications of text generation and related areas of natural language processing. These systems developed here help users in various domains to create high quality accurate and maintainable documents either automatically or semiautomatically.
These systems can be easily be inergrated with standard tools such as web browsers and word processors. They also developed a software called FoG (Forecast Generator) which generate a textual weather report from a map in both English and French. This has helped many forecasters to produce a weather forecast fast and easily since 1993. Automatic generation frees forecasters from the mechanical aspects of weather reports, allowing them to concentrate on aspects forcasting which most require human knowledge and intuition.
RealPro is CogenTex’s cross-platform, high performance syntantic realizer. It is designed to perform syntactic realization, i. e. the tranformation of abstract syntact specifications of natural language sentences (or phrases) to their corresponding surface forms at speed suitable for interactive processing. Currently CoGenTex is working on sevaral projects focus on diverse topics, from machine translation to computer assisted softawre engineering. NLP at MERL The Enlish writer’s assistant
Wring text in english presents a challenge to non-native speakers because of the difficulties in mastering English vocabulary, grammar and usage. Although most word-processing programs provide some kind of automatic grammar checking, these programmes ae not appropriate for helping non-native speakers to write English text as mistakes they make are different. MERL has developed a system specifically for non-native english speakers and in particular for japanese speakers. The system helps non-native speakers to compose English text while being tought about different aspects of Engish anguage has been built. The software developed here although in prottype stage is already very powerful, in addition to correcting the grammar the softwAare demonstrate many useful tool which help the user write Engish text. At MERL they are also working on developing computer suppoerted environment for collaborative learning, with a special focus on constructive andexpressive tools for distance larning, work and entertainment. Computerise and Online Dictionaries The idea here is very simple here any word can be looked in the dictionay just by typing and searching for it.
This give fast and very accurate results. Bank of English project by Cobuild is one exmample. Techniques There are sevaral main techniques used in analysing natural language processing. Some of them can be breafly described as follows. Pattern matching The idea here is an approach to natural language processing is to interpret input utterances as a whole father than builing up their interpretation by combining the structure and meaning of words or other lower level constituents.
That means the interpretations are obtained by matching patterns of words against the input utterance. For a deep level of analysis in pattern matching a large number of patterns are required even for a restricted domain. This problem can be ameliorated by hierarchical pattern matching in which the input is gradually canonicalized through pattern matching against subphrases. Another way to reduce the number of patterns is by matching with semantic primitives instead of words. Syntactically driven Parsing
Syntax means ways that words can fit together to form higher level units such as phrases, clauses and sentences. Therefore syntacticaly driven parsing means interpretation of larger groups of words are built up out of the interpretation of their syntacticconstituent words or phrases. In a way this is the opposite of pattern matching as here the interpretation of the input is done as a whole. Syntactic analyses are obtained by application of a grammar that determines what sentenses are legal in the language that is being parsed. Semantic Grammars
Natural language analysis based on semantic grammar is bit similar to systactically driven parsing except that in semantic grammar the catogaries used are defined semantically and syntactically. There here semantic grammar is also envolved. Case frame instantiation case frame instantiation is one of the major parsing techniques under active research today. The has some very useful computational properties such as its recursive nature and its ability to combine bottom-up recognition of key constituents with top-down instantiation of less structured constituents. Conclusions
Therefore it is clear that Natural Language Processing takes a very important roll in new machine human interfaces. When we look at some of the products that are based on technologies with NLP we can see that they are very advanced but very useful. But there are many limitations, requiring improvements and developement of NLP oriented systems. For example language we speak is highly ambiguous. This makes it very difficult understand and analyze. Also with so many languages spoken all over the world it is very difficult to design a system that is 100 % accurate.
These problems get more complicated when we think of different people speaking the same language with different styles. Therefore most of research on speech recognition is more concentrated on there areas. Information retrieval can be improved to give very accurate results for various searches. This will involve intelligence to find and sort all the results. So such intelligent systems are being experimented right now are we will be able to see improved applications of NLP in the near future.