DOCSLIB.ORG

Artificial Intelligence for Virtual Assistant

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Artificial Intelligence for Virtual Assistant Flyer Information Management and Governance Artificial Intelligence for Virtual Assistant IDOL for Virtual Assistant, is an AI (Artificial Intelligence) powered solution, that allows organizations to offer their customers and employees access to information and processes using an automated human-like operator which can engage in natural language dialogues. streamline the retrieval process, which allows such databases of information to allow natural To address an increasingly diverse information to be obtained in a more conve- language questions to receive direct factual nient and user-friendly fashion. Moreover, it answers. This is achieved by processing and range of information requests, should offer an interface to allow configura- understanding the question and mapping it to IDOL uses our unique Natural tion of that process to ensure existing human an appropriate structured query that will in turn Language Question Answering knowledge is allowed to create and train the respond with the desired answer. An example (NLQA) technology to allow natural system to answer questions optimally. In gen- of this would be databases of financial prices conversations between end users eral, there are three independent steps to ac- over time, allowing a query such as, “What was complish this. the EPS of HPQ in Q3 2016?” to receive the and a virtual assistant, to perform precise answer. This database of structure in- queries against: curated (FAQ) Answer Bank formation is generally known as the fact bank. responses (Answer Bank), data Many administrators of support or user-help tables and data services (Fact systems have an existing set of frequently In addition, the system should have the abil- Bank), and unstructured documents asked questions that human support agents ity to understand and extract information (for (Passage Extraction); and to are trained on, or the help pages are populated example names, numbers) from unstructured data such as free-form documents. It means follow configured processes to with. For example, if a user encounters a par- ticular problem on his mobile phone, then the that databases can be automatically populated facilitate knowledge discovery. manufacturer has established steps that the with a rich set of structured data from a corpus user should follow to try to correct the prob- of unstructured documents. For example, in- lem. The system should be loaded with these gesting a set of corporate annual reports could Question Answering answers to provide the best one when a given create tables of financial data so that a ques- The information retrieval process is at first question is asked or a particular problem is tion such as, “What was BP’s revenue in North glance well established—a user enters some encountered. In addition, the system’s natural America in the second half of 2015?” will return search terms, a Boolean expression, a docu- language processing (NLP) should ensure that the correct value pulled from such a table. ment, or a natural language query, and a set differently worded variants of the same ques- of relevant documents are returned—but in tions are all directed to the relevant answer. Passage Extraction many situations this workflow can be optimized This knowledge base is generally referred to In many cases, the information requested is further. If a user enters a direct question, for as the answer bank. simply not present in either an FAQ data set example, then it is generally more appropriate or a structured database, and an extended ap- to respond with a direct answer rather than a Fact Bank proach is required. It is at this point that the sys- document or documents that may contain the Users frequently have tables of information that tem must be able to process human information answer. The technology must be able to under- are commonly matched via structured (SQL) effectively. Once again, the question should be stand, process, and answer direct questions to queries. The system should be able to exploit processed to achieve a basic understanding Flyer Artificial Intelligence for Virtual Assistant of what is being requested and then, the sys- How do I turn o roaming on iPhone® 7 FAQ tem must form a query against its corpus to Answer bank support documents find the most relevant documents. The system Step by step instructions must process the selected portions of those documents to determine the short passage that it believes most accurately answers the What is HPQ’s Q3 FY16 EPS? Structured data original question. Statistical and probabilistic Fact bank from structured and $0.48 unstructured sources thresholding is used to determine whether an answer is relevant enough to answer the ques- tion, so that if a valid answer simply isn’t pres- What did critic X say about ent in the corpus then none is returned. Once movie Y? Passage Document extraction repositories again, the system must have the language-han- Quote from a TV interview dling capability to allow this process to work in any language. Figure 1. Question answering in natural language The above steps combine to form a powerful who will guide them to the correct solution or description. In many cases, the problem can data analytics system for processing ques- answer. However, large teams of human oper- be answered directly and a single answer or tions of any type. The typical workflow is to ators are costly and clients are often frustrated solution provided, but in many other cases, a first check whether the query is appropriate for by having to wait to talk to someone. So, how conversation between the system and the user question processing as many queries are not can we accelerate the right information to the is required to gather more information on how of this type. If it is, then it matches the ques- to diagnose and solve the problem. tion against the answer bank to see if a pre- right user in the right context? defined answer has been set. If none is found, it will then try the fact bank to see whether a A Natural Language Question Answering sys- This can be achieved by training the system precise answer is present. And again, if no an- tem can be extended to create a more natural with classes of problems and the type of infor- swer is found, then it will pass the query to the system of customer services. Rather than a mation that is needed in each case. The sys- system’s unstructured index to see if an ap- single search box, an IM-style interface allows tem will then respond with follow-up questions propriate passage can be found that answers a user to enter an initial question or problem until it believes it has enough knowledge to the question. Dialogue Systems I need the cheapest ticket from London to Boston For many users contacting a company for help • with a problem or an inquiry, a search box is in- • “Quantity?”, “Date and time?”, “Preferred airline?” sufficient. Even if the system is capable of -un • Ticket quantity, schedule, airline preference derstanding and processing direct questions, • Best options optimized for dierent parameters the user is often unable to accurately describe – Schedule, cost, mileage program the problem or need or fails to provide enough information. The natural response in such sit- uations is to want to talk to a human operator Figure 2. Dialogue system 2 Contact us at: www.microfocus.com Like what you read? Share it. suggest a solution. Here is a very simple exam- administrator to easily refine the answers’ avail- ple: user enters, “Cheapest ticket from London ability, accuracy and relevancy with a quick clicks. to Boston.” System asks, “Quantity, date and time, preferred airline?” User responds with Connecting Relevant Information answers. System presents best options opti- to Humans mized for different parameters: schedule, cost, Technology is only as good as the value real- mileage program. ized by its users. With the alluring promise of Big Data, comes some of the biggest barriers Simple Curation of Question/ to success—user adoption and productivity. Answer Pairs in Answer Bank True impact will come when humans can truly While it is natural to expect a manually inten- connect to technology. sive process for curating question/answer pairs (FAQs), IDOL comes with an intuitive GUI To learn more, please visit www.microfocus. tool to simplify such a process. This tool allows com/idol Figure 3. Question/Answer pairs in answer bank 161-000424-001 | H | 01/20 | © 2020 Micro Focus or one of its affiliates. Micro Focus and the Micro Focus logo, among others, are trademarks or registered trademarks of Micro Focus or its subsidiaries or affiliated companies in the United Kingdom, United States and other countries. All other marks are the property of their respective owners..
Load more

Recommended publications
  • Handling Ambiguity Problems of Natural Language Interface For
    IJCSI International Journal of Computer Science Issues, Vol. 9, Issue 3, No 3, May 2012 ISSN (Online): 1694-0814 www.IJCSI.org 17 Handling Ambiguity Problems of Natural Language InterfaceInterface for QuesQuesQuestQuestttionionionion Answering Omar Al-Harbi1, Shaidah Jusoh2, Norita Norwawi3 1 Faculty of Science and Technology, Islamic Science University of Malaysia, Malaysia 2 Faculty of Science & Information Technology, Zarqa University, Zarqa, Jordan 3 Faculty of Science and Technology, Islamic Science University of Malaysia, Malaysia documents processing, and answer processing [9], [10], Abstract and [11]. The Natural language question (NLQ) processing module is considered a fundamental component in the natural language In QA, a NLQ is the primary source through which a interface of a Question Answering (QA) system, and its quality search process is directed for answers. Therefore, an impacts the performance of the overall QA system. The most accurate analysis to the NLQ is required. One of the most difficult problem in developing a QA system is so hard to find an exact answer to the NLQ. One of the most challenging difficult problems in developing a QA system is that it is problems in returning answers is how to resolve lexical so hard to find an answer to a NLQ [22]. The main semantic ambiguity in the NLQs. Lexical semantic ambiguity reason is most QA systems ignore the semantic issue in may occurs when a user's NLQ contains words that have more the NLQ analysis [12], [2], [14], and [15]. To achieve a than one meaning. As a result, QA system performance can be better performance, the semantic information contained negatively affected by these ambiguous words.
    [Show full text]
  • Digital Workaholics:A Click Away
    GSJ: Volume 9, Issue 5, May 2021 ISSN 2320-9186 1479 GSJ: Volume 9, Issue 5, May 2021, Online: ISSN 2320-9186 www.globalscientificjournal.com DIGITAL WORKAHOLICS:A CLICK AWAY Tripti Srivastava Muskan Chauhan Rohit Pandey Galgotias University Galgotias University Galgotias University [email protected] muskanchauhansmile@g [email protected] m mail.com om ABSTRACT:-In today’s world, Artificial Intelligence (AI) has become an 1. INTRODUCTION integral part of human life. There are many applications of AI such as Chatbot, AI defines as those device that understands network security, complex problem there surroundings and took actions which solving, assistants, and many such. increase there chance to accomplish its Artificial Intelligence is designed to have outcomes. Artifical Intelligence used as “a cognitive intelligence which learns from system’s ability to precisely interpret its experience to take future decisions. A external data, to learn previous such data, virtual assistant is also an example of and to use these learnings to accomplish cognitive intelligence. Virtual assistant distinct outcomes and tasks through supple implies the AI operated program which adaptation.” Artificial Intelligence is the can assist you to reply to your query or developing branch of computer science. virtually do something for you. Currently, Having much more power and ability to the virtual assistant is used for personal develop the various application. AI implies and professional use. Most of the virtual the use of a different algorithm to solve assistant is device-dependent and is bind to various problems. The major application a single user or device. It recognizes only being Optical character recognition, one user.
    [Show full text]
  • Deep Almond: a Deep Learning-Based Virtual Assistant [Language-To-Code Synthesis of Trigger-Action Programs Using Seq2seq Neural Networks]
    Deep Almond: A Deep Learning-based Virtual Assistant [Language-to-code synthesis of Trigger-Action programs using Seq2Seq Neural Networks] Giovanni Campagna Rakesh Ramesh Computer Science Department Stanford University Stanford, CA 94305 {gcampagn, rakeshr1}@stanford.edu Abstract Virtual assistants are the cutting edge of end user interaction, thanks to endless set of capabilities across multiple services. The natural language techniques thus need to be evolved to match the level of power and sophistication that users ex- pect from virtual assistants. In this report we investigate an existing deep learning model for semantic parsing, and we apply it to the problem of converting nat- ural language to trigger-action programs for the Almond virtual assistant. We implement a one layer seq2seq model with attention layer, and experiment with grammar constraints and different RNN cells. We take advantage of its existing dataset and we experiment with different ways to extend the training set. Our parser shows mixed results on the different Almond test sets, performing better than the state of the art on synthetic benchmarks by about 10% but poorer on real- istic user data by about 15%. Furthermore, our parser is shown to be extensible to generalization, as well as or better than the current system employed by Almond. 1 Introduction Today, we can ask virtual assistants like Amazon Alexa, Apple’s Siri, Google Now to perform simple tasks like, “What’s the weather”, “Remind me to take pills in the morning”, etc. in natural language. The next evolution of natural language interaction with virtual assistants is in the form of task automation such as “turn on the air conditioner whenever the temperature rises above 30 degrees Celsius”, or “if there is motion on the security camera after 10pm, call Bob”.
    [Show full text]
  • Inside Chatbots – Answer Bot Vs. Personal Assistant Vs. Virtual Support Agent
    WHITE PAPER Inside Chatbots – Answer Bot vs. Personal Assistant vs. Virtual Support Agent Artificial intelligence (AI) has made huge strides in recent years and chatbots have become all the rage as they transform customer service. Terminology, such as answer bots, intelligent personal assistants, virtual assistants for business, and virtual support agents are used interchangeably, but are they the same thing? The concepts are similar, but each serves a different purpose, has specific capabilities, varying extensibility, and provides a range of business value. 2018 © Copyright ServiceAide, Inc. 1-650-206-8988 | www.serviceaide.com | [email protected] INSIDE CHATBOTS – ANSWER BOT VS. PERSONAL ASSISTANT VS. VIRTUAL SUPPORT AGENT WHITE PAPER Before we dive into each solution, a short technical primer is in order. A chatbot is an AI-based solution that uses natural language understanding to “understand” a user’s statement or request and map that to a specific intent. The ‘intent’ is equivalent to the intention or ‘the want’ of the user, such as ordering a pizza or booking a flight. Once the chatbot understands the intent of the user, it can carry out the corresponding task(s). To create a chatbot, someone (the developer or vendor) must determine the services that the ‘bot’ will provide and then collect the information to support requests for the services. The designer must train the chatbot on numerous speech patterns (called utterances) which cover various ways a user or customer might express intent. In this development stage, the developer defines the information required for a particular service (e.g. for a pizza order the chatbot will require the size of the pizza, crust type, and toppings).
    [Show full text]
  • Welsh Language Technology Action Plan Progress Report 2020 Welsh Language Technology Action Plan: Progress Report 2020
    Welsh language technology action plan Progress report 2020 Welsh language technology action plan: Progress report 2020 Audience All those interested in ensuring that the Welsh language thrives digitally. Overview This report reviews progress with work packages of the Welsh Government’s Welsh language technology action plan between its October 2018 publication and the end of 2020. The Welsh language technology action plan derives from the Welsh Government’s strategy Cymraeg 2050: A million Welsh speakers (2017). Its aim is to plan technological developments to ensure that the Welsh language can be used in a wide variety of contexts, be that by using voice, keyboard or other means of human-computer interaction. Action required For information. Further information Enquiries about this document should be directed to: Welsh Language Division Welsh Government Cathays Park Cardiff CF10 3NQ e-mail: [email protected] @cymraeg Facebook/Cymraeg Additional copies This document can be accessed from gov.wales Related documents Prosperity for All: the national strategy (2017); Education in Wales: Our national mission, Action plan 2017–21 (2017); Cymraeg 2050: A million Welsh speakers (2017); Cymraeg 2050: A million Welsh speakers, Work programme 2017–21 (2017); Welsh language technology action plan (2018); Welsh-language Technology and Digital Media Action Plan (2013); Technology, Websites and Software: Welsh Language Considerations (Welsh Language Commissioner, 2016) Mae’r ddogfen yma hefyd ar gael yn Gymraeg. This document is also available in Welsh.
    [Show full text]
  • The Question Answering System Using NLP and AI
    International Journal of Scientific & Engineering Research Volume 7, Issue 12, December-2016 ISSN 2229-5518 55 The Question Answering System Using NLP and AI Shivani Singh Nishtha Das Rachel Michael Dr. Poonam Tanwar Student, SCS, Student, SCS, Student, SCS, Associate Prof. , SCS Lingaya’s Lingaya’s Lingaya’s Lingaya’s University, University,Faridabad University,Faridabad University,Faridabad Faridabad Abstract: (ii) QA response with specific answer to a specific question The Paper aims at an intelligent learning system that will take instead of a list of documents. a text file as an input and gain knowledge from the given text. Thus using this knowledge our system will try to answer questions queried to it by the user. The main goal of the 1.1. APPROCHES in QA Question Answering system (QAS) is to encourage research into systems that return answers because ample number of There are three major approaches to Question Answering users prefer direct answers, and bring benefits of large-scale Systems: Linguistic Approach, Statistical Approach and evaluation to QA task. Pattern Matching Approach. Keywords: A. Linguistic Approach Question Answering System (QAS), Artificial Intelligence This approach understands natural language texts, linguistic (AI), Natural Language Processing (NLP) techniques such as tokenization, POS tagging and parsing.[1] These are applied to reconstruct questions into a correct 1. INTRODUCTION query that extracts the relevant answers from a structured IJSERdatabase. The questions handled by this approach are of Factoid type and have a deep semantic understanding. Question Answering (QA) is a research area that combines research from different fields, with a common subject, which B.
    [Show full text]
  • Quester: a Speech-Based Question Answering Support System for Oral Presentations Reza Asadi, Ha Trinh, Harriet J
    Quester: A Speech-Based Question Answering Support System for Oral Presentations Reza Asadi, Ha Trinh, Harriet J. Fell, Timothy W. Bickmore Northeastern University Boston, USA asadi, hatrinh, fell, [email protected] ABSTRACT good support for speakers who want to deliver more Current slideware, such as PowerPoint, reinforces the extemporaneous talks in which they dynamically adapt their delivery of linear oral presentations. In settings such as presentation to input or questions from the audience, question answering sessions or review lectures, more evolving audience needs, or other contextual factors such as extemporaneous and dynamic presentations are required. varying or indeterminate presentation time, real-time An intelligent system that can automatically identify and information, or more improvisational or experimental display the slides most related to the presenter’s speech, formats. At best, current slideware only provides simple allows for more speaker flexibility in sequencing their indexing mechanisms to let speakers hunt through their presentation. We present Quester, a system that enables fast slides for material to support their dynamically evolving access to relevant presentation content during a question speech, and speakers must perform this frantic search while answering session and supports nonlinear presentations led the audience is watching and waiting. by the speaker. Given the slides’ contents and notes, the system ranks presentation slides based on semantic Perhaps the most common situations in which speakers closeness to spoken utterances, displays the most related must provide such dynamic presentations are in Question slides, and highlights the corresponding content keywords and Answer (QA) sessions at the end of their prepared in slide notes. The design of our system was informed by presentations.
    [Show full text]
  • A Generator of Natural Language Semantic Parsers for Virtual
    Genie: A Generator of Natural Language Semantic Parsers for Virtual Assistant Commands Giovanni Campagna∗ Silei Xu∗ Mehrad Moradshahi Computer Science Department Computer Science Department Computer Science Department Stanford University Stanford University Stanford University Stanford, CA, USA Stanford, CA, USA Stanford, CA, USA [email protected] [email protected] [email protected] Richard Socher Monica S. Lam Salesforce, Inc. Computer Science Department Palo Alto, CA, USA Stanford University [email protected] Stanford, CA, USA [email protected] Abstract CCS Concepts • Human-centered computing → Per- To understand diverse natural language commands, virtual sonal digital assistants; • Computing methodologies assistants today are trained with numerous labor-intensive, → Natural language processing; • Software and its engi- manually annotated sentences. This paper presents a method- neering → Context specific languages. ology and the Genie toolkit that can handle new compound Keywords virtual assistants, semantic parsing, training commands with significantly less manual effort. data generation, data augmentation, data engineering We advocate formalizing the capability of virtual assistants with a Virtual Assistant Programming Language (VAPL) and ACM Reference Format: using a neural semantic parser to translate natural language Giovanni Campagna, Silei Xu, Mehrad Moradshahi, Richard Socher, into VAPL code. Genie needs only a small realistic set of input and Monica S. Lam. 2019. Genie: A Generator of Natural Lan- sentences for validating the neural model. Developers write guage Semantic Parsers for Virtual Assistant Commands. In Pro- templates to synthesize data; Genie uses crowdsourced para- ceedings of the 40th ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI ’19), June 22–26, 2019, phrases and data augmentation, along with the synthesized Phoenix, AZ, USA.
    [Show full text]
  • Ontology-Based Approach to Semantically Enhanced Question Answering for Closed Domain: a Review
    information Review Ontology-Based Approach to Semantically Enhanced Question Answering for Closed Domain: A Review Ammar Arbaaeen 1,∗ and Asadullah Shah 2 1 Department of Computer Science, Faculty of Information and Communication Technology, International Islamic University Malaysia, Kuala Lumpur 53100, Malaysia 2 Faculty of Information and Communication Technology, International Islamic University Malaysia, Kuala Lumpur 53100, Malaysia; [email protected] * Correspondence: [email protected] Abstract: For many users of natural language processing (NLP), it can be challenging to obtain concise, accurate and precise answers to a question. Systems such as question answering (QA) enable users to ask questions and receive feedback in the form of quick answers to questions posed in natural language, rather than in the form of lists of documents delivered by search engines. This task is challenging and involves complex semantic annotation and knowledge representation. This study reviews the literature detailing ontology-based methods that semantically enhance QA for a closed domain, by presenting a literature review of the relevant studies published between 2000 and 2020. The review reports that 83 of the 124 papers considered acknowledge the QA approach, and recommend its development and evaluation using different methods. These methods are evaluated according to accuracy, precision, and recall. An ontological approach to semantically enhancing QA is found to be adopted in a limited way, as many of the studies reviewed concentrated instead on Citation: Arbaaeen, A.; Shah, A. NLP and information retrieval (IR) processing. While the majority of the studies reviewed focus on Ontology-Based Approach to open domains, this study investigates the closed domain.
    [Show full text]
  • A Question-Driven News Chatbot
    What’s The Latest? A Question-driven News Chatbot Philippe Laban John Canny Marti A. Hearst UC Berkeley UC Berkeley UC Berkeley [email protected] [email protected] [email protected] Abstract a set of essential questions and link each question with content that answers it. The motivating idea This work describes an automatic news chat- is: two pieces of content are redundant if they an- bot that draws content from a diverse set of news articles and creates conversations with swer the same questions. As the user reads content, a user about the news. Key components of the system tracks which questions are answered the system include the automatic organization (directly or indirectly) with the content read so far, of news articles into topical chatrooms, inte- and which remain unanswered. We evaluate the gration of automatically generated questions system through a usability study. into the conversation, and a novel method for The remainder of this paper is structured as fol- choosing which questions to present which lows. Section2 describes the system and the con- avoids repetitive suggestions. We describe the algorithmic framework and present the results tent sources, Section3 describes the algorithm for of a usability study that shows that news read- keeping track of the conversation state, Section4 ers using the system successfully engage in provides the results of a usability study evaluation multi-turn conversations about specific news and Section5 presents relevant prior work. stories. The system is publicly available at https:// newslens.berkeley.edu/ and a demonstration 1 Introduction video is available at this link: https://www.
    [Show full text]
  • Natural Language Processing with Deep Learning Lecture Notes: Part Vii Question Answering 2 General QA Tasks
    CS224n: Natural Language Processing with Deep 1 Learning 1 Course Instructors: Christopher Lecture Notes: Part VII Manning, Richard Socher 2 Question Answering 2 Authors: Francois Chaubard, Richard Socher Winter 2019 1 Dynamic Memory Networks for Question Answering over Text and Images The idea of a QA system is to extract information (sometimes pas- sages, or spans of words) directly from documents, conversations, online searches, etc., that will meet user’s information needs. Rather than make the user read through an entire document, QA system prefers to give a short and concise answer. Nowadays, a QA system can combine very easily with other NLP systems like chatbots, and some QA systems even go beyond the search of text documents and can extract information from a collection of pictures. There are many types of questions, and the simplest of them is factoid question answering. It contains questions that look like “The symbol for mercuric oxide is?” “Which NFL team represented the AFC at Super Bowl 50?”. There are of course other types such as mathematical questions (“2+3=?”), logical questions that require extensive reasoning (and no background information). However, we can argue that the information-seeking factoid questions are the most common questions in people’s daily life. In fact, most of the NLP problems can be considered as a question- answering problem, the paradigm is simple: we issue a query, and the machine provides a response. By reading through a document, or a set of instructions, an intelligent system should be able to answer a wide variety of questions. We can ask the POS tags of a sentence, we can ask the system to respond in a different language.
    [Show full text]
  • Virtual Assistant Services Portfolio
    PORTFOLIO OF Virtual Assistant Services www.riojacinto.com WELCOME Hey there! My name is Rio Jacinto and I am a Virtual Assistant passionately helping busy business owners, women entrepreneurs and bloggers succeed by helping them with their website, social media accounts and others so they have more time to attract more ideal clients! I understand, You're here because you need an extra pair of hands to manage other business tasks that you're no longer have time to do. I get it, your main responsibility is to focus on building your business. I can help you with that. As a business owner you want to save some time to do more important roles in your business and I am here to help you and make that happen. Please check out my services and previous works in the next few pages. I can't wait to work with you! HOW CAN I HELP YOU MY SERVICES Creating Graphics using Canva Creating ebooks Personal errands (purchasing gifts for loved ones / family members online) Creating landing pages and sales pages FB Ads, Chatbots Hotel and Flight Booking Preparing Slideshows for Webinars or Presentations Recruitment and Outsourcing Management (source for other team members like writers) Social Media Management (Facebook, LinkedIn, Youtube, Pinterest, Instagram.) Creation of Basic Squarespace/Shopify Website Manage your Blog (Basic Squarespace/Shopify website Management) Filter and reply to comments on your blog Other virtual services, as available. PORTFOLIO FLYERS/MARKETING PORTFOLIO WEBSITE PORTFOLIO PINTEREST PORTFOLIO SOCIAL MEDIA GRAPHICS Facebook PORTFOLIO SOCIAL MEDIA GRAPHICS Instagram PORTFOLIO PRESENTATION WHITE PAPER RATES CHOOSE HERE FOR THE BEST PACKAGE THAT SUITS YOU Social Media Management Package starts from $150/mo *graphics creation/curation/scheduling/client support Pinterest Management Package starts from $50/wk *client provides stock photo or additional $2/ea.
    [Show full text]