File Information
File: 05-lr/acl_arc_1_sum/cleansed_text/xml_by_section/metho/00/w00-0404_metho.xml
Size: 30,166 bytes
Last Modified: 2025-10-06 14:07:19
<?xml version="1.0" standalone="yes"?> <Paper uid="W00-0404"> <Title>Extracting Key Paragraph based on Topic and Event Detection -- Towards Multi-Document Summarization</Title> <Section position="3" start_page="31" end_page="33" type="metho"> <SectionTitle> 2 Domain Dependency of Words </SectionTitle> <Paragraph position="0"> The domain dependency of words that how strongly a word features a given set of data (documents) contributes to event extraction, as we previously reported (Fukumoto et al.: 1997). In the study, we hypothesi~d that the articles from the Wall Street Journal corpus can be structured by three levels, i.e.</Paragraph> <Paragraph position="1"> Domain, Article and Paragraph. It'a word is nil event in a given article, it satisfies the two conditions: (1) The dispersion value of the word in the Paragraph level is smaller than that of the Art.iele, since the .word appears throughout paragr~q~hs in the Paragraph level rather than articles in the Article level. (2) The dispersion value of the word in the Article is smaller than that of the Domain, as the word appears across articles rather than domains.</Paragraph> <Paragraph position="2"> However, ~here are two problems to adapt it to multl-document summarization task. The first is that the method extracts only events in the document. Because the goal of the study is to summarize a single document, and thus there is no answer to the question of how to identi~' differences and similarities across documents. The second is that the performance of the method greatly depends on the structure of a given data itself. Like the Wall Street Journal corpus, (i) if a given data caal be structured by three levels, Paragraph, Article and Domain, each of which consists of several paragraphs, articles and domains, respectively, aaad (ii) if Domain consists of different subject domains, such as 'aerospace', 'environment' and 'stock market', the method can be done with satisfactoD' accuracy. However, there is no guarantee to make such an appropriate structure from a given set of documents in the multi-document summarization task.</Paragraph> <Paragraph position="3"> The purpose of this paper is to define domain dependency of words for a number of sample documents about the same topic, and thus for multi-document summarization task. Figure 4 illustrates the structure of broadcast news documents which have been developed by the TDT (Topic Detection and Tracking) Pilot Study (Allan and Carbonell, 1998). It consists of two levels, Paragraph and Document. In Document level, there is a small number of sample news documents about the same topic.</Paragraph> <Paragraph position="4"> These documents are arranged in chronological order such as, '(l-l) Quake collapses buildings in central ,Japan (Figure 2)', '(1-2) Two Americans known dead in Japan quake (Figure 1)' and '(1-3) gobe quake leaves questions about medical system (Figure 3)'. A particular document consists of several</Paragraph> <Paragraph position="6"> paragraphs. We call it Paragraph level. Let words within a document be an event, a topic, or among others (We call it n .qeneraZ word).</Paragraph> <Paragraph position="8"> Given the structure shown in Figure 4, how can we identi~&quot; every word in document (1-2) with an event, a topic or a general word? Our method assumes that aal event associated with a document appears across paragraphs, but a topic word does not. Then, we use domain dependency of words to extract event and topic words in document (1-2). Domain dependency of words is a measure showing how greatly each word features a given set of data.</Paragraph> <Paragraph position="9"> In Figure 4.. let 'C)', 'A' and 'x' denote a topicl an event and a general word in document (1-2), respectively. We recall the example shown in Figure 1.</Paragraph> <Paragraph position="10"> 'A', for instance, 'U.S.' appears across paragraphs.</Paragraph> <Paragraph position="11"> However, in the Document level, :A' frequently appears in document, (1-2) itself. On the basis of this example, we hypothesize that if word i is an event, it&quot;satisfies the following condition: \[1\] Word i greatly depends on a particular document in the Document level rather than a particular paragraph in the Paragraph. null Next, we turn to identi~&quot; the remains (words) wit.h a topic, or a general word. In Figure 5; a topic of documents (1-1) ~ (1-3), for instance, :Kobe' aPpears in a particular paragraph in each level of Paragraphl, Paragraph2 and Paragraph3. Here, (1-1), (12) and (1-3) corresponds to Paragraph1, Paragraph2 and Paragraph3, respectively. On the other hand, in Document level, a topic frequently appears acros.~ documents. Then: we hypothesize that if word i is a</Paragraph> <Paragraph position="13"> x: general word i leve iJ ! o x !~ j !C': .......</Paragraph> <Paragraph position="14"> l~igure 5: The structure of broadcast news documents (topic extraction) topic, it satisfies the following condition: \[2\] Word i greatly depends on a particular paragraph in each Paragraph level rather than a particular document in Document.</Paragraph> </Section> <Section position="4" start_page="33" end_page="33" type="metho"> <SectionTitle> 3 Topic and Event Extraction </SectionTitle> <Paragraph position="0"> We hypothesized that the domain dependency of words is a key clue to make a distinction between a topic and an event. This can be broken down into two observations: (i) whether a word appears across paragraphs (documents), (it) whether or not a word appears frequently. We represented the former by using dispersion value, and the latter by deviation value. Topic and event words are extracted by using these values.</Paragraph> <Paragraph position="1"> The first step to extract topic and event words is to assign weight to the individual word in a document. We applied TF*IDF to each level of the Document and Paragraph, i.e. Paragraphl, Paragraph2 and Paragraph3.</Paragraph> <Paragraph position="2"> N Wdit = TFdit * log Ndt (1) Wdit in formula (1) is TF*IDF of term t in the i-th document. In a similar way, Wpit denotes TF*IDF of the term t in the i-th paragraph. TFdit in (1) denotes term frequency of t in the i-th document. N is the number of documents and Ndt is the number of do(:uments where t occurs. The second step is to calculate domain dependency of words. We defined it by using formula (2) and (3).</Paragraph> <Paragraph position="4"> Formula (2) is dispersion value of term t in the level of Document which consists of m documents, and denotes how frequently t appears across documents.</Paragraph> <Paragraph position="5"> In a similar way, DispPt denotes dispersion of term t in the level of Paragraph. Formula (3) is the deviation value of t in the i-th document and denotes how frequently it appears in a particular document, the i-th document. Devpit is deviation of term t in the i-th paragraph. In (2) and (3), meant is the mean of the total TF*IDF values of term t in the level of Document.</Paragraph> <Paragraph position="6"> The last step is to extract a topic and an ever~t using fonmfla (2) and (3). We recall that if t is an event, it satisfies \[1\] described in section 2. This is shown by using formula (4) mad (5).</Paragraph> <Paragraph position="7"> DispPt < DispDt (4) for all Pi E di Devpjt < Devdit (5) Formula (4) shows that t frequently appears across paragraphs rather than documents. In formula (5), di is the i-th document and consists of the number of n paragraphs (see Figure 4). Pi is an element of di. (5) shows that t frequently appears in the i-th document di rather than paragraphs pj ( 1 < j < n). On the other hand: if t satisfies formula (6) and (7), then propose t as a topic.</Paragraph> <Paragraph position="8"> DispPt > DispDt (6) for all dl E D, Pit exists such that Devpjt >_ Devdlt (7) In formula (7), D consists of the number of rn docaments (see Figure 5). (7) denotes that t frequently appears in the particular paragraph pj rather than the document di which includes pj.</Paragraph> </Section> <Section position="5" start_page="33" end_page="33" type="metho"> <SectionTitle> 4 Key Paragraph Extraction </SectionTitle> <Paragraph position="0"> The summarization task in this paper is paragraph-based extraction (Stein et al., 1999). Basically, paragraphs which include not only event words but also topic words are considered to be significant paragraphs. The basic algorithm works as follows: 1. For each document: extract topic and event words.</Paragraph> <Paragraph position="1"> 2. Determine the paragraph weights for all paragraphs in the documents: (a) Compute the sum of topic weights over the total number of topic words for each paragraph. null (b) Compute the sum of event weights over the total number of event words for each paragraph. null A topic and an event weights are calculated by using Devdlt in formula (3). Here, t is a topic or an evcnt and i is the i-th document in the documents.</Paragraph> <Paragraph position="2"> (c) Compute the sum of (a) and (b) for each paragraph.</Paragraph> <Paragraph position="3"> 3. Sort the paragraphs t~ccording to their weights and extract the N highest weighted paragrai~hs in documents in order to yield summarization of the documents.</Paragraph> <Paragraph position="4"> 4. When their weights are the same, Compute the sum of all the topic and event word weights. Select a paragraph whose weight is higher than the others.</Paragraph> </Section> <Section position="6" start_page="33" end_page="75" type="metho"> <SectionTitle> 5 Experiments </SectionTitle> <Paragraph position="0"> Evaluation of extracting key paragraph based on multi-document is difficult. First, we have not found an existing collection of summaries of multiple documents. Second, the maamal effort needed to judge system output is far more extensive than for single document summarization. Consequently, we focused on the TDT1 corpus. This is because (i) events have been defined to support the TDT study effort, (ii) it was completely annotated with respect to these events (Allan and Carbonell, 1997). Therefore, we do not need the manual effort to collect documents which discuss about the target event.</Paragraph> <Paragraph position="1"> We report the results of three experiments. The first experiment, Event Extraction, is concerned with event extraction technique, ha the second experiment, Tracking Task, we applied the extracted topics to tracking task (Allan and Carbonell, 1998).</Paragraph> <Paragraph position="2"> The third experiment: Key Paragraph Extraction is conducted to evaluate how the extracted topic and event words can be used effectively to extract key paragraph.</Paragraph> <Section position="1" start_page="33" end_page="33" type="sub_section"> <SectionTitle> 5.1 Data </SectionTitle> <Paragraph position="0"> The TDT1 corpus comprises a set of documents (.15,863) that includes both newswire (Reuters) 7..965 and a manual transcription of the broadcast news speech (CNN) 7,898 documents. A set of 25 target events were defined 2 All documents were tagged by the tagger (Brill, 1992). %Ve used nouns in the documents.</Paragraph> <Paragraph position="2"/> </Section> <Section position="2" start_page="33" end_page="33" type="sub_section"> <SectionTitle> 5.2 Event Extraction </SectionTitle> <Paragraph position="0"> We collected 300 documents from the TDT1 corpus, each of which is mmolated with respect to one of 25 events.' The result is shown in Table 1.</Paragraph> <Paragraph position="1"> In Table 1, 'Event type' illustrates the target events defined by the TDT Pilot Study. 'Doe' denotes the number of documents. 'Rec' (Recall) is the immber of correct events divided by the total mnnber of events which are selected by a human, and 'Prec' (Precision) stands for the number of correctevents divided by the number of events which are selected by our method. The denominator 'Rec' is made by a hmnan judge. 'Accuracy' in Table 1 is the total average ratio.</Paragraph> <Paragraph position="2"> In Table 1, recall and precision values range from 55.0/47.0 to 83.3/84.2, the average being 71.0/72.2. The worst result of recall and precision was when event type was 'Serbs violate Bihac' (55.0/59.3). We currently hypothesize that this drop of accuracy is due to the fhct that some documents are against our assumption of an event. Examining the documents whose event type is 'Serbs violate Bihac', 3 ( one from CNN and two from Reuters).out of 16 documents has discussed the same event, i.e. 'Bosnian Muslim enclave hit by heavy shelling'. As a result, the event appears across these three documents* Future research will shed nmre light on that.</Paragraph> </Section> <Section position="3" start_page="33" end_page="33" type="sub_section"> <SectionTitle> 5.3 Tracking Task </SectionTitle> <Paragraph position="0"> Tracking task in the TDT project is starting from a few sample documents and finding all subsequent documents that discuss the same event (Allan and Carbonell, 1998), (Carbonell et al., 1999). The corpus is divided into two parts: training set and test set. Each of the documents is flagged as to whether it discusses the target event, and these flags ('YES', :'NO') axe the only information used for training the .system to correctly classiC&quot; the target event. We applied the extracted topic to the tracking task under a term vector For the results of topic extraction, all the documents that belong to the same topic are lmndled into a single document Stp and represent it by a term vector as follows:</Paragraph> <Paragraph position="2"> Let $1: --', S,, be all the other training documents (where m is the number of training documents which does not belong to the target event) and Sx be a test docmnent which should be classified as to whether or not it discusses the target event. 81, &quot;&quot; &quot;, Sm mid Sz are represented &quot; by term vectors as follows:</Paragraph> <Paragraph position="4"> appears ill S; and not, be a topic of Sip</Paragraph> <Paragraph position="6"> Compute the similarity between a training document and a test document Given a vector representation of documents SI, * * &quot;, Sin, Sty and Sx, a similarity between two documents Si (1 < i < m, tp) and the test document S~ would be obtained by using formula (8), i.e. the inner product of their normalized vectors.</Paragraph> <Paragraph position="8"> The greater the value of Sim(Si, S=) is, the more similar Si and S, are. If the similarity value between the test document Sx and the document Sip is largest among all the other pairs of documents, i.e. (&, S=).---, (S~, S=), Sx is judged to be a document that discusses the target event.</Paragraph> <Paragraph position="9"> We used the standard TDT evaluation measure In Table 3. 'Event' denotes event words in the first document in chronological order from A~ --- 4, and i the title of the document is 'Emergency Work Continues After Earthquake in Japan'. Table 3 clearly demonstrates that the criterion, domain dependency of-''words effectively employed.</Paragraph> <Paragraph position="10"> Figure 6 illustrates the DET (Detection Evaluation Tradeoff) curves for a sample event (event type. is 'Comet into Jupiter) runs at several values of Nt. \]~</Paragraph> <Paragraph position="12"> and 16. 'Miss' means Miss rate, which is the ratio of the doounents that were judged as YES but were not evahmted as YES for the run in question.</Paragraph> <Paragraph position="13"> 'F/A' shows false alarm rate and 'FI' is a measure that balances recall and precision. 'Rec' denotes the ratio of the documents judged YES that were also evaluated as YES, and 'Prec' is the percent of the documents that were evaluated as YES which correspond to documents actually judged as YES.</Paragraph> <Paragraph position="14"> Table 2 shows that more training data helps the performance, as the best result was when we used :Yt = 16.</Paragraph> <Paragraph position="15"> Table 3 illustrates the extracted topic and event words in a sample document. The topic is 'Kobe Japan quake' and the number of positive training documents is 4. 'Devpzt', 'Devd\]t', 'DispPt' and 'DispDt' denote values calculated by using formula (2) and (3).</Paragraph> <Paragraph position="16"> .ol .(m .o6 o.1 o2. o.5 1 g s lo '2o 4o $o 8o 90 Fatse Atarm p'rotm~Jity (in %) II Eigure 6: DET curve for a sample tracking runs * Overall, the curves also show that more training helps tile performance, while there is no significant B difference among -'Yt = 2, 4 and 8.</Paragraph> </Section> <Section position="4" start_page="33" end_page="36" type="sub_section"> <SectionTitle> 5.4 Key Paragraph Extraction </SectionTitle> <Paragraph position="0"> roll We used 4 different sets as a test data. Each set con- * sists of 2, 4.. 8 and 16 documents. For each set, we</Paragraph> <Paragraph position="2"/> </Section> <Section position="5" start_page="36" end_page="36" type="sub_section"> <SectionTitle> 5.2 Event Extraction </SectionTitle> <Paragraph position="0"> We collected 300 docmnents from the TDT1 corpus, each of which is annotated with respect to one of 25 events.' The result is shown in Table 1.</Paragraph> <Paragraph position="1"> In Table 1.. 'Event type' illustrates the target events defined by the TDT Pilot Study. ~Doc' denotes the number of documents. 'Rec' (Recall) is the nmnbet of correct events divided by the total number of events which are selected by a humaa, and :Pree ~ (Precision) stands for the number of correct-events divided by the number of events which are selected by our method. The denominator 'Rec: is made by a human judge. 'Accuracy' in Table 1 is the total average ratio.</Paragraph> <Paragraph position="2"> In Table 1, recall and precision values range, from 55.0/47.0 to 83.3/84.2, the average being 71.0/72.2.</Paragraph> <Paragraph position="3"> The worst result of recall and precision was when event type was 'Serbs violate Bihac' (55.0/59.3). We currently hypothesize that this drop of accuracy is due to the fact that some documents are against our assumption of an event. Examining the ctocuments whose event type is 'Serbs violate Bihac', 3 ( one from CNN and two from Reuters) out of 16 documents has discussed the same evefit, i.e. 'Bosnian Muslim enclave hit by heavy shelling'. As a result, the event appears across these three documents. Future research will shed more light on that.</Paragraph> </Section> <Section position="6" start_page="36" end_page="36" type="sub_section"> <SectionTitle> 5.3 Tracking Task </SectionTitle> <Paragraph position="0"> Tracking task in the TDT project is starting from a few sample documents and finding all subsequent documents that discuss the same event (Allan and Carbonell, 1998), (Carbonell et al., 1999). The corpus is divided into two parts: training set and test ~et. Each of the documents is flagged as to whether it discusses the target event, and these flags ('YES', 'NO') are the only information used tbr training the system to correctly classiC&quot; the target event. We applied the extracted topic to the tracking task under these conditions. The basic algorithm used in the * experiment is as follows: 1. Create a single document Stp and represent it as &quot;.a term vector For the results of topic extraction, all the documents that belong to the sanae topic are bundled into a single document S,p and represent it by a term vector as follows: Let $1, ---, S,,, be all the other training documents (where m is the number of training documents which does not belong to the target event) and Sx be a test document which should be classified as to whether or not it discusses the target event. $1, &quot;- -, Sm and Sx are represented &quot; by term vectors as follows:</Paragraph> <Paragraph position="2"> appears in S~ and not be a topic of ,5&quot;tp</Paragraph> <Paragraph position="4"> * S.t. t~j = f(t.r.j) ift~j ~ppears i, S, 0 otherwise 3. Compute the similarity between a training document and a test document Given a vector representation of documents SI, * .., S.,, Stp and S=; a similarity between two documents Si (1 < i < m, tp) mad the test document S= would be obtained by using formula (8), i.e. the inner product of their normalized vectors.</Paragraph> <Paragraph position="5"> Si * S= Sim(Si, S~) = I Si II S~ I (S) The greater the value of Sim(Si,S,) is, the more similar 5&quot;/ and Sz are. If the similarity value between the test document S, and the document Stp is largest among all the other pairs of documents, i,e. ($1, Sx), &quot; &quot;, (Sin, S=), S= is judged to be a document that discusses the target event.</Paragraph> <Paragraph position="6"> We used the standard TDT evaluation measure 3 and 16. 'Miss' means Miss rate, which is the ra- In Table 3, 'Event' denotes event words in the first rio of the documents that were, judged as YES but document in chronological order from .,X~ = 4, and not evaluated as YES for the run in question, the title of the document is 'Emergency Work Con- i were 'F/A' shows false Mann rate mad 'FI' is a measure tinues After Earthquake in Japan'. Table 3 clearly i that balances recall and precision. 'Rec' denotes the demonstrate~ that the criterion, domain dependency ratio of the documents judged YES that were also of words effectively employed, i evaluated as YES, and Tree' is the percent of the Figure 6 illustrates the DET (Detection Evalua- | documents that were evaluated as YES which corre- tion Tradeoff) curves for a sample event (event type spond to documents actually judged as YES. is 'Comet into Jupiter') runs at several values of Art. i Table 2 shows that more training data helps the performance, as the best result was when we used 9o ,.,.. ....... ---., , ...... , .... 'I -'Vt = 16. ~&quot; * q ..&quot; ~, ~ &quot;', .: * .. ,,~m~'~,~'----Table 3 illustrates the extracted topic and event E0 ~&quot; ...... &quot; .~&quot;'*'''&quot;=~''~'&quot;:&quot;*'&quot;'&quot;'&quot;'''~i .:',..: wt. &quot;.....: : ................ i e~ ....... * B words in a sample document. The topic is 'Kobe i i i iq i ~a'!4 ~- i i ~\[;\[: W Japem quake' and the m~mber of positive training e0 ~.4....i..-~...~....s....i-...i~:u...i...~,.4 ........ ~ ....... e,~, * i.: ~ &quot; ~ .'.~'~.'~ ;.~ &quot; ~. ~--','.: documents is 4. 'Devp\]t', 'Devd\]t', 'DispPt' and ~ l :. :. : : :.'-q: ;~ ~.~.: 1 : m~s-- ! 'DispDt' denote values calculated by using formula 4o : ~..~..~.~. : . . . : (2) and (3). : : : &quot; : : : : : : : ~ : ~..|.a. :i .: &quot; -: : -20 ~*..'2...,.t....'*--.:,*,*,?....;~::=~: : ; -- :*'*&quot;:'&quot;~&quot; ....... ! .......... ! ........... 9&quot;&quot;'&quot;*~ i</Paragraph> <Paragraph position="8"> Fat~ ~aan. Pr0ea~y fm ~) II Figure 6: DET curve for a sample tracking runs B 'Overall, the curves also show that more trailfing helps tile performance, while there is no significant difference anaong :Yt = 2, 4 and 8. il</Paragraph> </Section> <Section position="7" start_page="36" end_page="75" type="sub_section"> <SectionTitle> 5.4 Key Paragraph Extraction </SectionTitle> <Paragraph position="0"> We used 4 different sets as a test data. Each set con- I sists of 2, 4, 8 and 16 documents. For each set, we II</Paragraph> <Paragraph position="2"> extracted 10% and 20% of the full-documents para&quot;graph length (Jing et al., 1998). Table 4 illustrates the result.</Paragraph> <Paragraph position="3"> In Table 4, 'Num ~ denotes the number of documents in a set. 10 and 20deg~ indicate the extraction ratio. 'Para' denotes the number of par~]graphs exr.racted by a humaa~ judge, and 'Correct' shows the accuracy ot&quot; the method.</Paragraph> <Paragraph position="4"> The best result was 77.7% (the extraction ratio is 20% and the number of documents is 2).</Paragraph> <Paragraph position="5"> Wc now turn our attention to the main question: how was the contribution of making the distinction between a topic and an event for summarization task? Figure 7 illustrates the results of the methods which used (i) the extracted topic artd event words, i.e. our method, and (ii) only the extracted event&quot; words.</Paragraph> <Paragraph position="6"> In Figure 7, '(10%): and '(20%)' denote the extracted paragraph ratio. 'Event' is the result when we used only the extracted event words. Figure 7 shows that our method consistently outperforms the method which used only the extra,.ted events. To summarize the evaluation: ][: Event extraction effectively employed when each document discusses different subject about the same topic. This shows that the method will be applicable to other genres of corpora which consist of different subjects.</Paragraph> <Paragraph position="7"> 2. The result of tracking task (79.0% average recall and 86.6% average precision) is comparable to the existing tracking techniques which tested on the TDT1 corpus (Allan and Carbonell, 1998).</Paragraph> <Paragraph position="8"> 3. Distinction between a topic and an event improved the results of key paragraph extraction, as our method consistently outperforms the method which used only the extracted event words (see Figure 7).</Paragraph> </Section> </Section> <Section position="7" start_page="75" end_page="75" type="metho"> <SectionTitle> 6 Related Work </SectionTitle> <Paragraph position="0"> The majority of techniques for summarization fall within two broad categories: Those that rely on template instantiation and those that rely on passage extraction.</Paragraph> <Paragraph position="1"> Work in the former approach is the DARPA-sponsored TIPSTER program and, in particular, the message understanding conferences hag provided fertile groined for such work, by placing the emphasis of docunmnt analysis to the identification and extraction of certain core entities and facts in a document, while work on template-driven, knowledge.</Paragraph> <Paragraph position="2"> based summarization to date is hardly domain or genre-independent (Boguraev and Kennedy. 1997).</Paragraph> <Paragraph position="3"> The alternative approach largely escapes this constraint, by viewing the task as one of identi~,ing certain passages(typically sentences) which, by some metric, are deemed to be the most representative, of the document's content. A variety of approaches exist for determining the salient sentences in the text: statistical techniques based oll word distribution (Kupiec et al., 1995), (Zechner, 1996), (Salton et al., 1991), (Teufell and Moens, 1997), symbolic techniques based on discourse structure (Marcu, 1997) and semantic relations between words (Barzil~v and Elhadad, 1997). All of their results demonstrate that passage extraction techniques are a useful first step in document summarization, although most of them have focused on a single document.</Paragraph> <Paragraph position="4"> Some researchers have started to apply a single-document summarization technique to multidocument. Stein et. al. proposed a method for summarizing multi-document using single-document summarizer (Stralkowsik et al., 1998), (Stralkowski et al.. 1999). Their method first summarizes each document of multi-document, then groups the summaries in clusters and finally, orders these summaries in a logical way (Stein et al., 1999). Their technique seems sensible. However, as she admits, (i) the order the information should not only depend on topic covered, (ii) background information that helps clari~&quot; related information should be placed first. More seriously, as Barzilay and Mani claim, summarization of multiple documents requires information about similarities and differences across documents. Therefore it is difficult to identi~&quot; these information using a single-document summarizer technique (Mani and Bloedorn, 1997), (Barzilay et al., 1999).</Paragraph> <Paragraph position="5"> A method proposed by Mani et. al. deal with the problem, i.e. they tried to detect the similarities and differences in information content among documents (Mani and Bloedorn, 1997). They used a spreading activation algorithm and graph matching in order to identify similarities and differences across documents. The output is presented as a set of paragraphs with similar and unique words highlighted. However, if the same information is menNun: null &quot;tioned several times in different documents, much of the summary will be redundant.</Paragraph> <Paragraph position="6"> Allan et. al. also address the problem aald proposed a method for event tracking using common words and surprising features by supplementing the corpus statistics (Allan and Papka, 1998) (Papka et al., 1999). One of the purpose of this study is to make a distinction between an event aald an event class using surprising features. Here event class features are broad news areas such as politics, death, destruction and ~,'~fare. The idea is considered to be necessary to obtain higti accuracy, while Allan claims that the surprising words do not provide a broad enough coverage to capture all documents on the event.</Paragraph> <Paragraph position="7"> A more recent approach dealing with this problem is Barzilav et. al's approach (Barzilay et al., 1999). They used paraphrasing rules which are maaaually derived from the result of syntactic analysis to identify theme intersection and used language generation to reformulate them as a coherent, summary. While promising to obtain high accuracy: the result of summarization task has not been reported.</Paragraph> <Paragraph position="8"> Like Mani and Barzil~,'s techniques, our approach focuses on the problem that how to identi~&quot; differences and similarities across documents, rather than the problem that how to form the actual summar:,, (Sparck, 1993), (McKeown and Radev, 1995), (Radev and McKeown, 1998). However, while Barzilav's approach used paraphrasing rules to eliminate redmadancy in a summary, we proposed domain dependency of words to address robustness of the technique. null</Paragraph> </Section> class="xml-element"></Paper>