Anda di halaman 1dari 26

Penelusuran Informasi (Information Retrieval

)
Sumber: CS276: Information Retrieval and Web Search Pandu Nayak and Prabhakar Raghavan Term Vocabulary & Postings lists (Tokenisasi)

Penelusuran Informasi (Information Retrieval)

Ch. 1

Pertemuan sebelumnya:
 Struktur dari Inverted Indeks:
 Dictionary (Vocabulary) & Inverted List (Postings)

 Vocabulary urut berdasarkan term (kata)

 Untuk memproses Boolean Query:
 Melakukan interseksi (merging) secara linear

2

Penelusuran Informasi (Information Retrieval) Topik Pada Pertemuan Ini Tahapan dalam Membangun Indeks  Preprocessing untuk membentuk vocabulary  Documen  Tokenisasi (tokenization)  Kata (terms) apa saja yang dimasukkan dalam indeks  Inverted List (Postings)  Cara merge secara lebih cepat (faster merge) dengan cara skip lists  Query dalam bentuk kaliman (phrase) 3 .

Tokenizer Token stream Friends Romans Linguistic module Countrymen Modified tokens friend Indexer friend roman countryman 2 1 4 2 4 Inverted index roman countryman 13 16 . Romans.Introduction to Information Retrieval Diagram Proses Indexing Dokumen Friends. countrymen.

Penelusuran Informasi (Information Retrieval) Parsing Dokumen  Perhatikan terlebih dahulu format dokumen  pdf/word/excel/html ?  Ditulis dalam bahasa apa?  Format character set yang digunakan Bagaimana menentukan jawaban dari pertanyaan di atas? Observasi secara manual? Atau dilakukan secara otomatis menggunakan metode klasifikasi? 5 .

Penelusuran Informasi (Information Retrieval) Sec.1 Complications: Format/Language  Dokumen yang akan diindeks dapat berupa dokumen yang ditulis dalam beberapa bahasa  Sebuah indeks dapat mengandung kata dari beberapa bahasa  Karena sebuah dokumen dapat ditulis dalam beberapa bahasa  Contoh: Email dalam bahasa Inggris tetapi attacment dari email adalah dokumen yang ditulis dalam bahasa Jerman  Apakah unit dari sebuah dokumen?     Sebuah file? Sebuah email? Sebuah email dengan 5 attachments? Sekumpulan files (PPT atau halaman HTML)? 6 . 2.

Introduction to Information Retrieval TOKENS & TERMS (KATA) 7 .

Penelusuran Informasi (Information Retrieval) Sec.1 Tokenisasi (Tokenization)  Input: “Friends.2. Countrymen”  Output: Tokens  Friends  Romans  Countrymen  Jadi token adalah sederetan karakter (a sequence of characters) dalam dokumen  Setiap token menjadi kandidat dari elemen dalam indeks. tentunya setelah preprocessing 8 . 2. Romans.

Penelusuran Informasi (Information Retrieval) Sec. lower case?  San Francisco: satu token atau dua?  Bagaimana cara memutuskan bahwa SF adalah satu token? 9 .2.1 Tokenisasi (Tokenization)  Beberapa isu dalam tokenisasi:  Finland’s capital  Finland? Finlands? Finland’s?  Hewlett-Packard  Hewlett dan Packard sebagai dua token atau satu?  state-of-the-art: break up hyphenated sequence  co-education  lowercase. 2. lower-case.

2. 1991 12/3/91 No. Coba bayangkan bila ingin mencari baris dari error kode program melalui Sistem IR atau mencari nomor tertentu  Salah satu solusi adalah menggunakan mekanisme n-grams 10 .Penelusuran Informasi (Information Retrieval) Sec. 12.1 Angka (Numbers)     3/12/91 Mar. 2. B-52 Kode: 324a3df234cb23e Telepon: (0651) 234-2333  Biasanya angka memiliki space diantaranya  Sistem IR yang lama tidak mengindeks angka  Tapi angka itu penting.

2. tidak berhasil bila dicari via Google  Internationalization!  German noun compounds are not segmented  Lebensversicherungsgesellschaftsangestellter  ‘life insurance company employee’  German retrieval systems benefit greatly from a compound splitter module  Can give a 15% performance boost for German 11 .1 Tokenisasi: Isu dalam bahasa  French  L'ensemble  satu token atau dua?  L ? L’ ? Le ?  Want l’ensemble to match with un ensemble  Sampai tahun 2003.2.Penelusuran Informasi (Information Retrieval) Sec.

2.Penelusuran Informasi (Information Retrieval) Sec.000万円) Katakana Hiragana Kanji Romaji 12 .1 Tokenisasi: Isu dalam bahasa  Chinese and Japanese:  莎拉波娃现在居住在美国东南部的佛罗里达。  Not always guaranteed a unique tokenization  Further complicated in Japanese: Dates/amounts in multiple formats フォーチュン500社は情報不足のため時間あた$500K(約6. 2.

1 Tokenisasi: Isu dalam bahasa  Tulisan Arab ditulis dari kanan ke kiri tetapi untuk angka dibaca dari kiri ke kanan  Words are separated. but letter forms within a word form complex ligatures  ← → ←→ ← start  ‘Algeria achieved its independence in 1962 after 132 years of French occupation.Penelusuran Informasi (Information Retrieval) Sec.’ 13 .2. 2.

Penelusuran Informasi (Information Retrieval) Sec. to.2 Stop words  Menggunakan stop list. and. be  Jumlahnya cukup banyak: ~30% dari semua kata dalam corpus  Trend: stopword tidak diikutkan:  Hemat indeks dan dapat memperkecil ukuran indeks walaupun dikompres  Query optimisasi menjadi lebih baik  Tapi perlu juga memperhatikan Query sbb:  Judul film: “King of Denmark”  Judul Lagu: “Let it be”. kata-kata yang sering muncul (tetapi kurang penting) dapat dikeluarkan dari indeks:  Secara semantic mereka tidak penting: the. 2. “To be or not to be”  Relational query: “flights to London” 14 . a.2.

Penelusuran Informasi (Information Retrieval) Sec.S. antidiscriminatory  antidiscriminatory 15 . e.A.S.2.A.. USA  USA  deleting hyphens to form a term  anti-discriminatory.g.3 Normalisasi Kata (terms)  Kata harus dinormalisasidalam in indexed text as well as query words into the same form  We want to match U. 2. which is an entry in our IR system dictionary  We most commonly implicitly define equivalence classes of terms by.. and USA  Result is terms: a term is a (normalized) word type.  deleting periods to form a term  U.

Tubingen  Tubingen 16 . French résumé vs. Tübingen. German: Tuebingen vs.2. users often may not type them  Often best to normalize to a de-accented term  Tuebingen.g.  Umlauts: e. Tübingen  Should be equivalent  Most important criterion:  How are your users like to write their queries for these words?  Even in languages that standardly have accents.g...Penelusuran Informasi (Information Retrieval) Sec. 2. resume.3 Normalization: other languages  Accents: e.

Chinese characters  Tokenization and normalization may depend on the language and so is intertwined with language detection Morgen will ich in MIT … Is this German “mit”?  Crucial: Need to “normalize” indexed text as well as query terms into the same form 17 .3 Normalization: other languages  Normalization of things like date forms  7月30日 vs. 7/30  Japanese use of kana vs.2.Penelusuran Informasi (Information Retrieval) Sec. 2.

A. 2.g.3 Case folding  Reduce all letters to lower case  exception: upper case in mid-sentence?  e. 18 .2. General Motors  Fed vs. sail  Often best to lower case everything.Penelusuran Informasi (Information Retrieval) Sec. since users will use lowercase regardless of ‘correct’ capitalization…  Google example:  Query C..  #1 result was for “cat” (well.T. Lolcats) not Caterpillar Inc. fed  SAIL vs.

windows.2.3 Normalization to terms  An alternative to equivalence classing is to do asymmetric expansion  An example of where this may be useful  Enter: window  Enter: windows  Enter: Windows Search: window. window Search: Windows  Potentially more powerful.Penelusuran Informasi (Information Retrieval) Sec. 2. windows Search: Windows. but less efficient 19 .

by hand-constructed equivalence classes  car = automobile color = colour  We can rewrite to form equivalence-class terms  When the document contains automobile. index it under carautomobile (and vice-versa)  Or we can expand a query  When the query contains automobile.g. which forms equivalence classes of words based on phonetic heuristics  More in lectures 3 and 9 20 ..Penelusuran Informasi (Information Retrieval) Thesauri and soundex  Do we handle synonyms and homonyms?  E. look under car as well  What about spelling mistakes?  One approach is soundex.

Penelusuran Informasi (Information Retrieval) Sec.4 Lemmatization  Reduce inflectional/variant forms to base form  E. is  be  car.g.  am.. car's.2. 2. cars'  car  the boy's cars are different colors  the boy car be different color  Lemmatization implies doing “proper” reduction to dictionary headword form 21 . cars. are.

4 Stemming  Reduce terms to their “roots” before indexing  “Stemming” suggest crude affix chopping  language dependent  e. 2.2. automate(s). automation all reduced to automat. for exampl compress and compress ar both accept as equival to compress 22 .Penelusuran Informasi (Information Retrieval) Sec.. for example compressed and compression are both accepted as equivalent to compress.g. automatic.

select the one that applies to the longest suffix.4 Porter’s algorithm  Commonest algorithm for stemming English  Results suggest it’s at least as good as other stemming options  Conventions + 5 phases of reductions  phases applied sequentially  each phase consists of a set of commands  sample convention: Of the rules in a compound command. 23 . 2.Penelusuran Informasi (Information Retrieval) Sec.2.

Penelusuran Informasi (Information Retrieval) Sec. 2.2.4 Typical rules in Porter     sses  ss ies  i ational  ate tional  tion  Rules sensitive to the measure of words  (m>1) EMENT →  replacement → replac  cement → cement 24 .

German.ac. 2.2.g. Helps recall but harms precision  operative (dentistry) ⇒ oper  operational (research) ⇒ oper  operating (systems) ⇒ oper  Definitely useful for Spanish.uk/computing/research/stemming/general/lovins. Lovins stemmer  http://www. longest suffix removal (about 250 rules)  Full morphological analysis – at most modest benefits for retrieval  Do stemming and other normalizations help?  English: very mixed results.Penelusuran Informasi (Information Retrieval) Sec.comp.4 Other stemmers  Other stemmers exist. Finnish.lancs. e. …  30% performance gains for Finnish! 25 ..htm  Single-pass.

Can we do better? Yes (if index isn’t changing too fast).3 Recall basic merge  Walk through the two postings simultaneously.Penelusuran Informasi (Information Retrieval) Sec. 2. in time linear in the total number of postings entries 2 2 8 1 4 2 8 3 41 8 48 11 64 17 128 21 Brutus 31 Caesar If the list lengths are m and n. the merge takes O(m+n) operations. 26 .