} END TERM EXAMINATION [MAY. 2019] SIXTH SEMESTER [BCA] ARTIFICIAL INTELLIGENCE [BCA-312] imo: $ hrs, Note: Attempt five qu one questi M.M.:75 tions in all including question no. 1 which is compulsory. Select on from each unit, Q.1, Attempt any five of the following: (5 x 5.= 25) Q.1. GY Differentiate between A* AND AO* search algorithm. Which one is bottor? Ans, Heuristic methods offer good strategies but fail to deseribe how the shortest nee (o 8 goal should be estimated. ‘The A* algorithm is a specialization of best- first search, The A® algorithm is a specialization of best-first search. A* algovith Step 1, Place the starting node s on open. Stop 2. Ifopen is empty, stop and return failure. Step 3. Remove from open the node n that has smallest value of f(n). If the node is agoal node, return success and stop. Otherwise, Step 4. successor 1’, and n, generating all of its successors n’ and place n on closed. For every n’ is not already on open or closed attach a back-pointer to n, compute Pw) and place it on open, Step 5. Each n’ that is arleady on open or closed should be attached to back-pointers which roflect the lowest g*(v”) path, If n’ was on closed and its pointer was changed, remove it and place it on open. Stop 6. Return to Step 2 AO* Algorith Step 1. Pl ce the start node s on open Step 2. Using the search tree constructed thus far, compute the most promising | lution Tree Ty, | Step 8, Select a node n that is both on open and a part of Ty. Remove n from open . nd place it on closed. wea Step 4, If'n is a terminal goal node, label n as solved. 3 : / Bip 5. Ifnis not a’solvable node, label nas unsolvable. If the start node is labelled | Sutsolvable, exit with failure. |, Step 6, Otherwise, expand node n generating all of | “stsor node that contains more than one subproblem, 1 eer tote vidual subproblems. Attach to each newly generated no A ode _Meeessor. Compute the east estimate h* for each newly genera | Step 7, ‘ts successors. For each such generate their successors to i ‘eturn to step 2. . oo be shown that AO* will always find a mini i a itioncy depends on how closely h* approximates h. ike A* mum-cost solution tree. Like A*, Scanned with CamScannerneater, Artificial Intelligence Sixth So 2-201 arning and its significance. Qui. (b) Explain Rote Le ‘Ans, Rote Learning, Rote Learning is basien Saving knowledge so it can be used again. «+ Retrieval in the only problem. + No repeated computation, inf Ansimple example of rote learning is caching + Store computed values (or large piece of data) «+ Recall this information when required by computation. 1y memorisation. nce or query is necessary. nificant time savings ean be achieved. + Many Al programs (as well as’ more general ones) have used y for learning: caching very effectively. Memorisation is a key neces: « [tin a basic necessity for any intelligent. program — is it a separate learning process’? + Memorisation can be a complex subject — how best to store knowledge? n employed rote learning (it also used parameter Samuel's Cheekers prog adjustment which will be discussed shortly). * Aminimax search was used to explore the game tree. * ‘Time constraints do not permit complete searches. * IL records board positions and scores at search ends. * Now if the same board position arises later in the game the stored value can be recalled and the end effect is that more deeper arched have occurred. Rote learning, is basically a simple process. However it does illustrate some issues that are relevant to more complex learning issues. Organisation «Access of the stored value must be faster than it would be to recompute it. Methods such as hashing, indexing and sorting can be employed to enable this. E.g Samuel's program indexed board positions by noting the number of pieces. Generalisatio +The number of potentially stored objects can be very large. We may need to cralise some information to make the problem manageable. 4,4 Samuel's program stored game positions only for white to move. Also rotations ee along diagonals are combined. Stability of the Environment ‘ - Rote learning is not very effective in a rapidly changing environment. If the environment does change then we must detect and record exactly what has changed — the frame problem, Q.1. (c) What is constraint satisfaction? Discuss with example Ans, In artificial intelligence and operations research, constraint satisfaction '5 the process of finding a solution to a set of constraints that impose conditions that the ables must satisfy . A solution is therefore a set of values for the variables that satinficw all constraints that is, a point in the feasible region. _ Tho techniques used in constraint satisfaction depend on the kind of co! boing, considered, Often used are constraints on a finite domain, to the poin! nstraints that Scanned with CamScanneri LP. University-[BCA]~Akash Books 2019. | constraint satisfaction problems are typicully identified with problems based on constraints on a finite domain, Such problems are usually solved via search, in particular a form of backtracking or local search, C sation are other methods used on such problems; most of them are incomplete in general, that is, they may solve the problem or prove it unsatisfiable, but not always, Constraint propagation methods are also used in conjunction with se: 1 riven problem simpler to solve, Other considered kinds of consi re on real or rational numbers; solving, problems on these constraints'is done via variable climination or the simplex algorithm. Constraint satisfaction originated in the field of artificial intelligence in the 1970s. During the 1980s and 1990s, embedding of constraints inlo a programming language was developed. Languages often used for constraint programming are Prolog and C++. Q.1. (d) Describe mapping between facts and representation. Ans. In all variety of knowledge representations, we deal with two kinds of entities. A. Facts: ‘Truths in some relevant world. ‘These are the th B. Representations of faets in some chosen formal actually be able to manipulate. h to make gs we want to represent. n. These are things we will One way to think of structuring these entities is at two levels: (a) the knowledge level, at which facts are des ribed, and (b) the symbol level, at which representations of objects at the knowledge level are defined in terms of symbols that can be manipulated by programs. The facts and representations are linked with two-way mappings. This link is ealled representation mappings. The forward representation mapping maps from facts to representations. The backward representation mapping goes the other way, from representations to facts, One common representation is natural language (particularly English) sentences. Regardless of the representation for facts we use in a program, we may also need to be concernied with an English representation of those facts in order to facilitate getting information into and out of the system. We need mapping functions from English sentences to the representation we actually use and from it back to sentences. Q.1. (e) Differentiate between forward and backward reasoning. Ans. In Artificial intelligence, the purpose of the search is to find the path through a problem space. There are two ways to pursue such a search that are forward and backward reasoning. The significant difference between both of them is that forward teasoning starts with the initial data towards the goal. Conversely, backward reasoning } Works in opposite fashion where the purpose is to determine the initial facts and information with the help of the given results. Comparison Chart [Basis for Comparison] Forward Reasoning Backward Reasoning | [Pasi Data-driven Goal driven | Petins with Objective |New Data Uncertain conclusion | iefind the Conclusion that must follow | Facts to support the conclusions| 4 lye fapproach Opportunistic Conservative. = Incipient to consequence _| Consequence to incipient 91.0 Define AI. List various applications of Al in its related fields. Ans. Refer Q.8, (b) Kind ‘Term Examination 2018. (Page No. 7-2018) Scanned with CamScannersto sixth Semester, Attificial Intelligence UNIT Qa. Whatis productionsystem? Discuss the production rules and solution for the problem of "Missionaries amd Can nibnl”. (b) Rnd Term Examination 2017, (Page No, 7-2017) ] MISSIONARIES AND CANNIBALS PROBLEM. | ‘Three missionaries and three cannibals stand on the west bank of a river. Nearby ‘ean accommodate either one or two persons, How can everyone cross cannibals never outnumber missionaries on s ever outmumber the missionaries on either shore or on the boat, then those mi will be eaten. Before beginning a search, we - rast decide on a representation for the problem. We can use Side; OmOc; Oo represent the start state. This denotes that three missionaries (3m) and three cannibals (3c) are ene he West bank, no missionaries (Om) nor cannibals (0c) are on the East bank, The final 0 indicates that the boat is on the West bank, whereas 1 in this position specifies that the boat is on the East bank, (Acomputer program written to solve this problem might represent the start-state ‘as 39000: the representation as given in this text [3m3c; OmOe; 0] adds clarity for peaders.) The goal state is accordingly represented by the string OmOc; m3c; 1 Moves Wil be tried in the order m, me, 2¢, and c, which represent that a single missionary, @ tnissionary and cannibal, (wo cannibals, and a lone cannibal ave to cross the river (note Din wee not considered for use here). The direction of travel is made obvious by the boat's position. To ensure that our notation is clear, we provide a bfs solution in Figure below, which has been extended to two levels. is. a row boat t! to the east bank in such a either side of the river? Ifthe Jiwest bank; east bank: boat on west x loop x occours Isewhere Scanned with CamScannerLP. University-{BCA]-Akash Books 2019-5 Note that no move was pursued that, would lead to an unsafe state on either bank. Nota also that repeated states are to be pruned. Moves are tried in the order m, me, 2¢ and. ___ Q.2. (b) Discuss “Hill climbing” search algorithm along with its various issues. (6) Ans. Refer Q.1. (4) End Term Examination 2018. (Page No. 1-2018) and Refer Q.3. (a) End Term Examination 2017. (Page No. 8-2017) Q.3 (a) What are Agents in AI? How Agents work to import intelligence to a system? Classify the different type of agents and briefly discuss their properties. 6.5) Ans. Artificial intelligence is defined as a study of rational agents. A rational agent could be anything which makes decisions, as a person, firm, machine, or software. It carries out an action with the best outcome after considering past and current precepts (agent’s perceptual inputs at a given instance). An AI system is composed of an agent and its environment. The agents act in their environment. The environment may contain other agents. An agent is anything that can be viewed as: * Perceiving its environment through sensors and * Acting upon that environment through actuators Note: Every agent can perceive its own actions (but not always the effects) Sensors Actuators we should be familiar with inery that the agent executes ic ear, a camera, a PC. agent function is a perceived till date) to re of Intelligent Agents, i To understand the struc! Taner Architecture and Agent Program. hinery on Itiea device with sensors and actuators, for example: arbor Agent program is an implementation of an agent function. Ap map from the percept sequence (Iistory of all that an ag ’ an action, ; = Architecture + Agent Program = a erpants, voxel Agent Archiectre A Ss agent has Keystrokes, Me - ee Gao or at act as sensors and displays on the screen, packages Packets acting as actuators. lege Human agent has eyes, es, mouth, and other body par dhands, i Jas sensors ant or organs which acl ears, and other ts acting as actuators. Scanned with CamScanner6-2019 Sixth Semester, Artificial Intelligence A Robotic agent has Cameras and various motors acting as actuators. Abilities infrared range finders which act.as sensory g nd Goals/Preferences Agent Prior Knowledge Actions Observations * Past Experiences Types of Agents Agents can be grouped into four classes based on their degree of perceived intelligence and capability: + Simple Reflex Agents * Model-Based Reflex Agents Goal-Based Agents Utility-Based Agents * Learning Agent Simple reflex agents Simple reflex agents ignore the rest of the percept history and act only on the basis of the current percept. Percept: history is the history of all that an agent has perceived till date. The agent function is based on the condition-action rule. A condition-action rule is a rule that maps a state i.e, condition to an action. If the condition is true, then the action is taken, else not. This agent function only succeeds when the environment is filly observable. For simple reflex agents operating in partially observable environments, infinite loops are often unavoidable. It may be possible to escape from infinite loops if the agent can randomize its actions. Problems with Simple reflex agents are : * Very limited intelligence. : + No knowledge of non-perceptual parts of state. * Usually too big to generate and store. * Ifthere occurs any change in the environment, be updated. then the collection of rules needs to auouN9s}AUS Actuators Scanned with CamScanner‘ IP. University-[BCA|-Akash Books 2019-7 Model-based reflex agents: It wor i joddel-b orks by findin, i | mca Amon rea ang ne pin mh | yuseof adel aboxt the world. The agent has to keep track of internal state whichis | adjusted by each perce pt and buat depends on the percept history. The current state is a medetnee intains some kind of structure describit the world which eannot be seen. Updating the state requires information coecte + how the world evolves in-dependently from the agent, and : j | + how the agent actions affects the world. What the world] is like now ‘What my actions do | How the world quouruostaug, What action T [should do new] v Actuators [Condition-netion ral Goal-based agents: These kind of agents take decision based on how far they are currently from their goal(description of desirable situations). Their every action is intended to reduce its distance from the goal. This allows the agent a way to choose dmong multiple possibilities, selecting the one which reaches a goal state. The knowledge that supports its decisions is represented explicitly and can be modified, which makes these agents more flexible. They usually require search and planning, The goal-based agent's behavior can easily be changed. Precepts ee : evolves What the world] jae the world evolves lke cow What my actions do v Whavit will be | 7 like if 1 do action 4} | quowuosy Aug a ' ‘What action I should do new Scanned with CamScanner8-2019 Sixth Semester, Artificial Intelligence i developed having their end uses a, ility-based agents: The agents which are 1 : ie auap eee failed utility based agents. When there are multiple possibl, on then to decide which one is best, utility-based agents are used. They choose alternatives, then ees ili +h state. Someti i ‘d on a preference (utility) for eacl , desi goal poopie We may look for a quicker, safer, cheaper trip to reach a destination, Agent happiness should be taken into consideration. Utility deseri bes Ee happy” the agent is. Because of the uncertainty in the world, a utility agent chooses the action that Inaximives the expected utility. A utility function maps a state onto a real number which describes the associated degree of happiness. Sensors ¥ $770 0F Precepts i t State What the world is like now What my actions do v [How the world evolves What it will be like if 1 do action Al | How happy I will Utility be in such a state ns What action T should do new v Actuators quowuosmaug Learning Agent Alearning agent in Al is the type of agent which can learn from its past experiences or it has learning capabilities. It starts to act with basic knowledge and then able to act and adapt automatically through learning. Alearning agent has mainly four conceptual components, which are: 1. Learning element :It is responsible for making improvements by learning from the environment 2. Critic: Learning element takes feedback from critic which describes how well the agent is doing with respect to a fixed performance standard. 8.Performance element: It is responsile for selecting external action 4, Problem Generator: This component is responsible for suggesting actions that will lead to new and informative experiences, Scanned with CamScannerLP. University{BCA]-Akash Books 2019-9 ‘AGENT Performance standard 1 ¥ Pe ts Critic = ¢ Sensors <}——— feedback | | x Changes Learning [| Performance element | knowledge| element Learning goals, Problem ‘Experiments generator Effectors. Actions Q.3. (b) What is state space search? What is Heuristic Search? (6) Ans. State Space Search: Refer Q.2. (a) End Term Examination 2017. (Page No. 6-2017) A heuristic is a method that + might not always find the best solution + But is guaranteed to find a good solution in reasonable tune. « By sacrificing completeness it increases efficiency. + Useful in solving tough problems which + Could not be solved any other way. — Solutions take an infinite time or very long time to compute. ‘The classic example of heuristic search methods is the travelling salesman. problem. Heuristic Search methods + Generate and Test Algorithm + Hill climbing . + Best First Search Problem Reduction (AO* Algorithm) UNIT: problem solving? Explain with @) quations between symbolic ywed to occur In are considered der variables, ess is Q44. (a) What is Unification algorithm in example. Ans. Unification is an algorithmic process of solving © ‘apressons, Depending on which expressions (also called terms) are allo @nequation set (also called unification problem), and which expressions “qual, several frameworks of unification are distinguished. If higher-ort oe that is, variables representing functions, are ‘allowed in an expression te process Sled higher orior cnification, otherwise first-order uniticablon: Hfgaataion rewired “omake both sides of each equation literally ‘equal, the process is called $9 Scanned with CamScanner10-2019 Sixth Semester, Artificial Intelligence free unification, otherwise semantic or equational unification, or E-unification, oy tnifiention modulo theory. ; If two literals are to unify, then first, their predicate symbols must match; if they do not, then these two literals cannot be unified. For example: Kite, is_ flying (X) and ‘Irying_.to_fly_Kite (Y) cannot be unified. [f the predicate symbols match, then you check the arguments one pair ata time. If the first matches, then continue with the second, and so on. (4, (b) IMustrate the use of First Order Logic to represent Knowledge in detail along with example. (6.5) Ans. First-order logic is symbolized reasoning in which each sentence, or statement, down into a subject and a predicate. The predicate modifies or defines the :# of the subject. In first-order logic, a predicate can only refer to a single subject. First-order logic is also known as first-order predicate calculus or first-order functional calculus, | A sentence in first-order logic is written in the form Px or P(x), where Pis the predicate anil x is the subject, represented as a variable. Complete sentences are logically combined and manipulated according to the same rules as those used in Boolean algebra. In first-order logic, a sentence can be structured using the universal quantifier (aymbolized ‘/) or the exist represented by x. Hirst Order Logic * The prepositional logic only deals, with the facts that may be true or false. tial quantifier (3). Consider a subject that is a variable * The first. order logic assumes that the world contains objects, relations and functions. Syntax for first order logic: In prepositional logic, every expression is a sentence that represents a fact, * Hirst order logic includes the sentences along with terms which can represent the objects. “C quantifie tant symbols, variables and function symbols are used to build terms, while and predicate symbols are used to build the sentences. Syntax: | Constants ABCs j Functions Size, Color ' Variable xa ‘Terms Constant, variable or function(Term..) Predicates ‘true, False Quantifiers va Atomic sentences: Predicate, Predicate(Term,.... Term = Term Sentences ~ Sentence, Sentence v Sentence, Sentence » Sentence, Sentence <> Sentence. Sentence <> Sentence, Quantifier variable,.. Sentence Scanned with CamScannerAkash Books 2019-11 Semantics: Lets understand rem: ye ta re Has Size (Elephant, Big) Size OF (Elephant) = Big Lets introduce a new syntax: a object Elephant is big, which Elephants in general. So let's represent th statement a: + All thi + For all things x, for which x i ence “Elephants are big’. gs that are Elephant. * For all things x. if x is a Elephant, then xis big. Finally the FOL will be written as. ¥ x Elephant (x) <> Bigtx) Knowledge Engineering of FOL ‘The knowledge engi objects and relationships. The following steps are used to develop the knowledge base: 1. Identify the problem or ta: 2, Assemble the relevant knowledge to the given problem or task. 3. Decide on a vocabulary or predicates, functions and constants 4, Encode the general knowledge about the domian and a descripton of the specific problem instance. 5. Apply queri 6. Finally, debug the knowledge ba: Q.5 (a) Write following sentences into formulae in predicate logic and a nt the important must know about the domain to repres to the inference procedure and get the answers into clausial form. (i) John links all food. (ii) Apple is food iii) Chicken are food (iy) Anything anyone eats and isn’t killed by is food. (v) Bill eats peanuts and is not killed. (vi) Sue eats peanuts and is not killed. Ans, Refer Q.4, End Term Examination 201 Q5. (b) Prove that “john likes peanut Resolution method with above sentences. Ans. To prove that: Like (John, Peanuts) Conversion of above wis into clause form: (1: ~ Food (x) v Likes (John, %) C2: Food (Apples) 4: Food (Chicken) 7. (Page No. 9-2017) 5” using backward chaining and Scanned with CamScannerSixth Somostor, Artificial Intelligence: BO OAs WN ys Bast yy) > Killed (x) > Pood (y) eats i, 9). = Kitlod x) v Pood (9) = ill, Peanuts) ive (RADY OW ~ Killed (HD) 1 (Sue,.v) is sot of support) [o] ; / ow xi? Poamuts cy 7 Pome) Meant = y/P Peanuts, ats (x, Peanuts) v Killed (x) wit\ / iia) Killed (BILD But from sub clause C5(b) we have Alive (Bill), i.e. Bill is alive, So occurred. Therefore, our assumption that John does not like Peanuts is false, Hence we ntradiction has can say that: Likes (John, Peanuts) is true. Q.6. (a) What is the role of semantic and syntactic processing in natural language processing? (65) | Ans, Refer of Q.7. (a) End Term Examination 2018, (Page No. 13-2018) Q.6. (b) Discuss Discourse and Pragmatic processing with the help of an example. (6) Ans. + Pragmatics: It deals with using and understanding sentences in different situations and how the interpretation of the sentence is affected. Pragmatic analysis helps users to discover this intended effect by applying a set of rules that characterize cooperative dialogues. E.g., “close the window?" should bo interpreted as a.request instead of an order, ___* Discourse: It deals with how the immediately preceding sentence ean affect the interpretation oft the next sentence, It means a sense of the context. The meaning of AY Fae Sentence which depends upon that sentence. It also considers the meaning of the vowing sentence. For example, the word “that” in the sentence “He wanted that depends upon the prior discourse context, Q.7. Discuss the followi (a Q.7. (a) Explanation based learning. Ans. Explanation-Based Learning (EBL) is « form of machine learning (st ization? be ae a very strong, or even perfect, domain theory in order to make gener: or form concepts from training examples, Good application domain for an EBL language processing (NLP). Scanned with CamScannerLP. University-[BCAl-Akash Books 2019-13 Anexample of BBL using a perfect domain theory is a program that learns to play chess through example. A specific chess position that contains an important Pisture such as “Forced loss of black queen in two moves” includes many irrelevant features, such as the specific scattering of pawns on the board. EBL: can takea single | training example and determine what the relevant features are in order to form a generalization. |" BBL system works by finding a way to deduce each training example from the system's existing database of domain theory. Havinga short proof ofthe training example | extends the domain-theory database, enabling the EBL syster) find and classify future gxamples that are similar to the training example very quickly. The main drawback of the mothod-_the eost of applying the learned proof macros, as these become numerous — | was analyzed by Minton. | BBL software takes four inputs: | « A hypothesis space (the set of all possible conclusions) | + Adomain theory (axioms about ¢ domain of interest) | straining examples (specific facts that rule out some possible hypothesis) + Operationality criteria (criteria for determining which features in the domain are leficientiy recognizable, e.g, which features are directly detectable usiag sensors) Q.7. (b) Learning from example induction ‘Ans. * Induction learning is carried out in the « In this learning process, a general rule is in observed instance. * However, class definitions can be method, Q:7. (c) Learning by taking advice ‘Ans. « This type is the easiest and simple way of learning, « In this type af learning, a programmer writes a program to gi) tN" instructions to perform task to the computer. Once itis learned (i.e. programmed), the system will be able to do new things. « Also, there can be several internet ete. } «However, this typ { + As the stored knowl: form, the reliability of the | basis of supervised learning. duced by the system from a set of constructed with the help of a classification 1 sources for taking advice such as humans (experts), of inference than rote learning. ansformed into an operational _ n into consideration. .¢ of learning has a more necessity ledge in knowledge base gets tr knowledge source is always take UNIT-IV g with its detailed architecture. ‘What is Q.8. (a) Define Expert System alonj (6.5) the role of expert system shell? a Ans, Refer Q.9, End Term Examination 2018. (Page No. 17-2018) Expert System Shells An Expert system shell is basic components of expert SYS! contains the r t environment. It a software developmen! comethod for ae A shell i associated with a proseripe itlding applications by configuring and instantiating these components scat Shell components and description: The generic components A a lnowledge acquisition, the knowledge Base, the reasoning, the expansion a org iterfaee soe chive below. ‘The knowledge base and reasoning enéi ‘omponents. Scanned with CamScanner14-2019 Sixth Semester, Artificial Intelligence Expert System Shell E x [inference | ; j Mechanism xplanation Pp Reascning with 3 e Uncertainty [Subsystem . t Knowledge Engineer Q.8. (b) How is an expert system built? Discuss one example. 6 Ans. Refer Q.9. (a) of End Term Examination 2017. (Page No. 16-2017) Q.9. (a) Compare procedural programming language and Al programming language. 6) Ans, Procedural Programming: Procedural Programming can be defined asa programming model which is derived from structured programming, based upon the concept of calling procedure. Procedures, also known as routines, subroutines or functions, simply consist of a series of computational steps to be carried out. During a program's execution, any given procedure might be called at any point, including by other procedures or itself. Languages used in Procedural Programming: FORTRAN, ALGOL, COBOL, BASIC, Pascal and C. Al programming language: ‘The primary goals of Al include deduction and reasoning, knowledge representation, planning, natural language processing (NLP), learning, perception, and the ability to manipulate and move objects. Long-term goals of Al research include achieving Creativity, Social Intelligence, and General (human level) Intelligence. Al has heavily influenced different sectors that we may not recognize, Ray Kurzweil says “Many thousands of AI applications are deeply embedded in the infrastructure of every industry.” John McCarthy, one of the founders of AI, once said that “as soon asit works, no one calls it AI anymore.” Broadly, ATis classified into the following: . Machine Learning: Inspired by the strudy of pattern recognition, machine learning enables software algorithm that are self-teaching and can make predictioss about data. _ Nerual Network: Taking cues from biological neural networks like the hums? jmeves System, artifical neural networks process a large number of novel and wnidt® inputs, collecting and rationalising it into ‘knowledge’. ANNs are perfect for the big 4" World with its myriad file and data types. . Deep Learing: After synthesising inputs of all types, deep learning i Just like yout can carry out a complex abstract task (like making coffee) one-off inputs, deep learning tries to model high-level abstractions about W set of inputs might combine to de pee 8. (b) Write LISP program to find sum and average of any four 14" Ans-Refer Q.8. (b) End Term Examination 2017. (Page No. 16-2017) sthe nests? from count ynat a give? Scanned with CamScann er