Full Text

It searches a potential redex depth-first and from left to right: datatype found = VAL of value | POTRED of p ot en t ia l_ re d ex (* term → found *) fun s ea rc h _t er m_ n eg ( VAR x ) = VAL ( NEGVAR_nf x ) | s ea r ch _t er m _n eg ( NEG t ) = POTRED ( PR_NEG t ) | s ea r ch _t er m _n eg ( CONJ ( t1 , t2 )) = POTRED ( PR_CONJ ( t1 , t2 )) | s ea r ch _t er m _n eg ( DISJ ( t1 , t2 )) = POTRED ( PR_DISJ ( t1 , t2 )) (* term → found *) fun search_term ( VAR x ) = VAL ( POSVAR_nf x ) | search_term ( NEG t ) = s ea r ch _t er m _n eg t | search_term ( CONJ ( t1 , t2 )) = ( case search_term t1 of ( VAL v1 ) ⇒ ( case search_term t2 of ( VAL v2 ) ⇒ VAL ( CONJ_nf ( v1 , v2 )) | ( POTRED pr ) ⇒ POTRED pr ) | ( POTRED pr ) ⇒ POTRED pr ) | search_term ( DISJ ( t1 , t2 )) = ( case search_term t1 of ( VAL v1 ) ⇒ ( case search_term t2 of ( VAL v2 ) ⇒ VAL ( DISJ_nf ( v1 , v2 )) | ( POTRED pr ) ⇒ POTRED pr ) | ( POTRED pr ) ⇒ POTRED pr ) (* term → found *) fun search t = search_term t When a negation is encountered, the auxiliary function search_ term_neg is called to decide whether this negation is a value or a potential redex. 2.1.2 From searching to decomposing Let us transform the search function of Section 2.1.1 into a decomposition function for the reduction semantics of Section 2.2.

CPS transformation: The search function is CPS-transformed by naming its intermediate results, sequentializing their computation, and introducing an extra functional argument, the continuation, that maps an intermediate result to a final answer: (* term × ( found → α) → α *) fun s ea rc h _t er m_ n eg ( VAR x , k ) = k ( VAL ( NEGVAR_nf x )) | s ea r ch _t er m _n eg ( NEG t , k ) = k ( POTRED ( PR_NEG t )) | s ea r ch _t er m _n eg ( CONJ ( t1 , t2 ) , k ) = k ( POTRED ( PR_CONJ ( t1 , t2 ))) | s ea r ch _t er m _n eg ( DISJ ( t1 , t2 ) , k ) = k ( POTRED ( PR_DISJ ( t1 , t2 ))) (* term × ( found → α) → α *) fun search_term ( VAR x , k ) = k ( VAL ( POSVAR_nf x )) | search_term ( NEG t , k ) = search_term_neg (t , k ) | search_term ( CONJ ( t1 , t2 ) , k ) = search_term ( t1 , fn ( VAL v1 ) ⇒ search_term ( t2 , fn ( VAL v2 ) ⇒ k ( VAL ( CONJ_nf ( v1 , v2 ))) | ( POTRED pr ) ⇒ k ( POTRED pr )) | ( POTRED pr ) ⇒ k ( POTRED pr )) | search_term ( DISJ ( t1 , t2 ) , k ) = search_term ( t1 , fn ( VAL v1 ) ⇒ search_term ( t2 , fn ( VAL v2 ) ⇒ k ( VAL ( DISJ_nf ( v1 , v2 ))) | ( POTRED pr ) ⇒ k ( POTRED pr )) | ( POTRED pr ) ⇒ k ( POTRED pr )) (* term → found *) fun search t = search_term (t , fn f ⇒ f ) Simplifying the CPS-transformed search: The search is completed as soon as a potential redex is found. It can thus be simplified by only applying the continuation when a value is found: (* term × ( value → found ) → found *) fun s ea rc h _t er m_neg ( VAR x , k ) = k ( NEGVAR_nf x ) | s ea r ch _t erm_neg ( NEG t , k ) = POTRED ( PR_NEG t ) | s ea r ch _t erm_neg ( CONJ ( t1 , t2 ) , k ) = POTRED ( PR_CONJ ( t1 , t2 )) | s ea r ch _t erm_neg ( DISJ ( t1 , t2 ) , k ) = POTRED ( PR_DISJ ( t1 , t2 )) (* term × ( value → found ) → found *) fun search_term ( VAR x , k ) = k ( POSVAR_nf x ) | search_term ( NEG t , k ) = s ea r ch _t erm_neg (t , k ) | search_term ( CONJ ( t1 , t2 ) , k ) = search_term ( t1 , fn v1 ⇒ search_term ( t2 , fn v2 ⇒ k ( CONJ_nf ( v1 , v2 )))) | search_term ( DISJ ( t1 , t2 ) , k ) = search_term ( t1 , fn v1 ⇒ search_term ( t2 , fn v2 ⇒ k ( DISJ_nf ( v1 , v2 )))) (* term → found *) fun search t = search_term (t , fn v ⇒ VAL v ) Potential redexes are now returned directly and the VAL constructor is relegated to the initial continuation.

Full Text

A battery is a cell which can provide electricity. ...... 2) Reduced pattern instances (capitalized tags are chunks and syntactic classes): NN , < Search_Term> , DT$ NN according to < Search_Term> , DT$ NNP known as < Search_Term> , VB < Search_Term> BE$ DT$ ...... 3) Soft patterns based on the instances: ...... < Search_Term> ...... NN 0.12 , 0.11 , 0.40 DT$ 0.2 according 0.03 to 0.03 BE$ 0.2 VB 0.1 known 0.09 as 0.20 In Table 1, centroid words are those words that are highly correlated to the search term, as judged by mutual information.

Full Text

We'll use a single hardware platform with multiple cores and a Linux operating system (see Table 1). auto kern_start( map.link< raft::out >( kernel::make< filereader >( file, offset ), kernel::make< search< ahocorasick /** or boyermoore **/ > >( search_term ) ) ); map.link< raft::out >( &(kern_start.dst), kernel::make< write_each< match_t > >( std::back_inserter( total_hits ) ) ); Figure 9: Implementation of the string matching application topology using RaftLib.